Methods, systems, apparatuses, and computer programs for processing tomographic images

BACKGROUND

Field

0001The present application relates generally to obtaining tomographic images in a dental environment, and, more particularly, to methods, systems, apparatuses, and computer programs for processing tomographic images.

Description of Related Art

0002X-ray radiography can be performed by positioning an x-ray source on one side of an object (e.g., a patient or a portion thereof) and causing the x-ray source to emit x-rays through the object and toward an x-ray detector (e.g., radiographic film, an electronic digital detector, or a photostimulable phosphor plate) located on the other side of the object. As the x-rays pass through the object from the x-ray source, their energies are absorbed to varying degrees depending on the composition of the object, and x-rays arriving at the x-ray detector form a two-dimensional (2D) x-ray image (also known as a radiograph) based on the cumulative absorption through the object. This process is explained further in reference to <figref idref="f0070">FIG. 60A-FIG. 60C</figref>.

0003<figref idref="f0070">FIG. 60A</figref> shows a patient and an x-ray device <b>6000</b> positioned for obtaining an occlusal image of the mandible. A sensor or x-ray film (not shown) is disposed inside the patient's mouth. An exposure is recorded and an x-ray image is subsequently developed, as shown in <figref idref="f0070">FIG. 60B</figref>. In traditional x-ray imaging, the sensor (or film) records the intensity of the x-rays incident thereon, which are reduced (or attenuated) by the matter which lies in their respective paths. The recorded intensity represents that total attenuation of the x-ray through the image volume, as illustrated in <figref idref="f0070">FIG. 60C</figref>.

0004In <figref idref="f0070">FIG. 60C</figref> two x-rays (ray A and ray B) are incident on two adjacent sensor elements <b>6008</b> and <b>6010</b> of an x-ray detector. Ray A travels through two blocks of identical material, block <b>6002</b> and block <b>6004,</b> each with an attenuation factor of 50%. Ray B travels through block <b>6006</b> which also has an attenuation factor of 50%. The intensity of Ray A recorded by sensor element <b>6008</b> is 25% of the original intensity. The intensity of Ray B recorded by sensor element <b>6010</b> is 50% of the original intensity. In a traditional x-ray image the recorded intensities are represented by light/dark regions. The lighter regions correspond to areas of greater x-ray attenuation and the darker regions correspond to areas of less, if any, x-ray attenuation. Thus, a two-dimensional projection image produced by sensor elements <b>6008</b> and <b>6010</b> will have a lighter region corresponding to sensor element <b>6008</b> and a darker region corresponding to sensor element <b>6010.</b> However, from such a two-dimensional projection image it cannot be determined that there were two blocks of material <b>(6002</b> and <b>6004)</b> at different positions in the path of Ray A, as opposed to a single block of material which caused the same amount of x-ray attenuation. In other words, the traditional x-ray image contains no depth information. As such, overlapping objects may easily obscure one another and reduce the diagnostic usefulness of the projection image.

0005Computed tomography (CT) and cone beam computed tomography (CBCT) have been used to acquire three-dimensional data about a patient, which includes depth information. The three-dimensional data can be presented on a display screen for clinician review as a 3D rendering or as a stack of parallel 2D tomographic image slices. Each slice represents a cross-section of the patient's anatomy at a specified depth. While CT and CBCT machines may produce a stack of parallel 2D tomographic image slices, these machines carry a high cost of ownership, may be too large for use in chair-side imaging, and expose patients to a relatively high dose of x-rays.

0006Tomosynthesis is an emerging imaging modality that provides three-dimensional information about a patient in the form of tomographic image slices reconstructed from images taken of the patient with an x-ray source from multiple perspectives within a scan angle smaller than that of CT or CBCT (e.g., ±20°, compared with at least 180° in CBCT). Compared to CT or CBCT, tomosynthesis exposes patients to a lower x-ray dosage, acquires images faster, and may be less expensive.

0007Typically, diagnosis using tomosynthesis is performed by assembling a tomosynthesis stack of two-dimensional image slices that represent cross-sectional views through the patient's anatomy. A tomosynthesis stack may contain tens of tomosynthesis image slices. Clinicians locate features of interest within the patient's anatomy by evaluating image slices one at a time, either by manually flipping through sequential slices or by viewing the image slices as a cine loop, which are time-consuming processes. It may also be difficult to visually grasp aspects of anatomy in a proper or useful context from the two-dimensional images. Also, whether the tomographic images slices are acquired by CT, CBCT, or tomosynthesis, their usefulness for diagnosis and treatment is generally tied to their fidelity and quality.

0008Quality may be affected by image artifacts. Tomosynthesis datasets typically have less information than full CBCT imaging datasets due to the smaller scan angle, which may introduce distortions into the image slices in the form of artifacts. The extent of the distortions depends on the type of object imaged. For example, intraoral tomosynthesis imaging can exhibit significant artifacts because structures within the oral cavity are generally dense and radiopaque. Still further, spatial instability in the geometry of the tomosynthesis system and/or the object can result in misaligned projection images which can degrade the quality and spatial resolution of the reconstructed tomosynthesis image slices. Spatial instability may arise from intentional or unintentional motion of the patient, the x-ray source, the x-ray detector, or a combination thereof. It may therefore be desirable to diminish one or more of these limitations.

0009<patcit id="pcit0001" dnum="US6196715B1"><text>US 6,196,715B1</text></patcit> discloses an x-ray diagnostic system preferable to two dimensional x-ray detection. Reference is further made to <nplcit id="ncit0001" npl-type="s"><text>Ewa Pietka et al, "Computer-Assisted Bone Age Assessment: Image Preprocessing and Epiphyseal/Metaphyseal ROI extraction", IEEE TRANSACTIONS ON MEDICAL IMAGING, vol. 20, no. 8, August 2001, ISSN 0278-0062</text></nplcit>. <patcit id="pcit0002" dnum="US2014169650A1"><text>US2014169650A1</text></patcit> discloses a radiation tomographic image generating method, and radiation tomographic image generating program. <patcit id="pcit0003" dnum="WO2013056733A1"><text>WO2013056733A1</text></patcit> discloses out of plane artifact reduction in digital breast tomosynthesis and ct.

SUMMARY

0010One or more the above limitations can be diminished by the method as defined in claim 1, and the system as defined in claim 7, and the storage medium as defined in claim 8.

0011In one embodiment, a method of identifying a tomographic image of a plurality of tomographic images is provided. Information specifying a region of interest in at least one of a plurality of projection images or in at least one of a plurality of tomographic images reconstructed from the plurality of projection images is received. A tomographic image of the plurality of tomographic images is identified. The identified tomographic image is in greater focus in an area corresponding to the region of interest than others of the plurality of tomographic images.

0012In another embodiment, an apparatus for identifying a tomographic image from a plurality of tomographic images. The apparatus includes a processor and a memory storing at least one control program. The processor and memory are operable to: receive information specifying a region of interest in at least one of a plurality of projection images or in at least one of a plurality of tomographic images reconstructed from a plurality of projection images, and identify a tomographic image of the plurality of tomographic images. The tomographic image is in greater focus in an area corresponding to the region of interest than others of the plurality of tomographic images.

0013In a further embodiment, a non-transitory computer-readable storage medium storing a program which, when executed by a computer system, causes the computer system to perform a method. The method includes receiving information specifying a region of interest in at least one of a plurality of projection images or in at least one of a plurality of tomographic images reconstructed from the plurality of projection images, and identifying a tomographic image of the plurality of tomographic images. The identified tomographic image is in greater focus in an area corresponding to the region of interest than others of the plurality of tomographic images.

0014In still another embodiment, a method for generating clinical information is provided. Information indicating at least one clinical aspect of an object is received. Clinical information of interest relating to the at least one clinical aspect is generated from a plurality of projection images. At least one of the steps is performed by a processor in conjunction with a memory.

0015In still a further embodiment, an apparatus for generating clinical information. The apparatus includes a processor and a memory storing at least one control program. The processor and the memory are operable to: receive information indicating at least one clinical aspect of an object, and generate, from a plurality of projection images, clinical information of interest relating to the at least one clinical aspect.

0016In yet another embodiment, a non-transitory computer readable storage medium storing a program which, when executed by a computer system, causes the computer system to perform a method. The method includes receiving information indicating at least one clinical aspect of an object, and generating, from a plurality of projection images, clinical information of interest relating to the at least one clinical aspect.

BRIEF DESCRIPTION OF THE DRAWINGS

0017The teachings claimed and/or described herein are further described in terms of exemplary embodiments. These exemplary embodiments are described in detail with reference to the drawings. These embodiments are non-limiting exemplary embodiments, in which like reference numerals represent similar structures throughout the several views of the drawings, and wherein: <ul id="ul0001" list-style="none" compact="compact"><li><figref idref="f0001">FIG. 1A</figref> is a system block diagram of a tomosynthesis system according to one example embodiment herein.</li><li><figref idref="f0002">FIG. 1B</figref> illustrates an example of a linear scan path used by the tomosynthesis system according to an example embodiment herein.</li><li><figref idref="f0003">FIG. 1C</figref> illustrates an example of a curved scan path used by the tomosynthesis system according to an example embodiment herein.</li><li><figref idref="f0004">FIG. 1D</figref> illustrates an example of a circular scan path used by the tomosynthesis system according to an example embodiment herein.</li><li><figref idref="f0005">FIG. 1E</figref> illustrates an example of shadow casting from an orthogonal projection angle.</li><li><figref idref="f0006">FIG. 1F</figref> illustrates an example of shadow casting from a non-orthogonal projection angle and the parallax induced in the image of the objects.</li><li><figref idref="f0007">FIG. 2A</figref> illustrates a block diagram of an example computer system of the tomosynthesis system shown in <figref idref="f0001">FIG. 1A</figref>.</li><li><figref idref="f0008">FIG. 2B</figref> is a flowchart illustrating a procedure for generating clinical information from a tomosynthesis dataset according to an example aspect herein.</li><li><figref idref="f0009">FIG. 3</figref> is a flowchart illustrating a procedure for identifying high-focus images within a tomosynthesis dataset according to an example embodiment herein.</li><li><figref idref="f0010">FIG. 4</figref> illustrates an example orthogonal projection image of a tooth and a region of interest indication thereon.</li><li><figref idref="f0011">FIG. 5</figref> illustrates an example focus profile within the region of interest of the tooth of <figref idref="f0010">FIG. 4</figref>.</li><li><figref idref="f0012">FIG. 6</figref> illustrates an example high-focus tomosynthesis image slice of the tooth of <figref idref="f0010">FIG. 4</figref>.</li><li><figref idref="f0012">FIG. 7</figref> illustrates another example high-focus tomosynthesis image slice of the tooth of <figref idref="f0010">FIG. 4</figref>.</li><li><figref idref="f0013">FIG. 8</figref> illustrates the example orthogonal projection image of <figref idref="f0010">FIG. 4</figref> with a portion of <figref idref="f0012">FIG. 6</figref> overlaid thereon within the region of interest.</li><li><figref idref="f0013">FIG. 9</figref> illustrates the example orthogonal projection image of <figref idref="f0010">FIG. 4</figref> with a portion of <figref idref="f0012">FIG. 7</figref> overlaid thereon within the region of interest.</li><li><figref idref="f0014">FIG. 10</figref> illustrates an example user interface that displays high-focus tomosynthesis image slices according to an example embodiment herein.</li><li><figref idref="f0015">FIG. 11</figref> illustrates an example tomosynthesis image slice with a nerve canal indicated.</li><li><figref idref="f0016">FIG. 12</figref> illustrates an example tomosynthesis image slice with a sinus cavity indicated.</li><li><figref idref="f0017">FIG. 13A</figref> illustrates an example 2D radiograph.</li><li><figref idref="f0017">FIG. 13B</figref> illustrates an example tomosynthesis image slice of the anatomy of <figref idref="f0017">FIG. 13A</figref>, wherein the nasal cavity is indicated.</li><li><figref idref="f0017">FIG. 13C</figref> illustrates another example tomosynthesis image slice of the anatomy of <figref idref="f0017">FIG. 13A</figref>, wherein the nasal cavity is indicated.</li><li><figref idref="f0018">FIG. 14A</figref> illustrates an example 2D radiograph.</li><li><figref idref="f0018">FIG. 14B</figref> illustrates an example tomosynthesis image slice of the teeth of <figref idref="f0018">FIG. 14A</figref>, wherein a crack is visible.</li><li><figref idref="f0018">FIG. 14C</figref> illustrates another example tomosynthesis image slice of the teeth of <figref idref="f0018">FIG. 14A</figref>, wherein a crack is visible.</li><li><figref idref="f0019">FIG. 15</figref> illustrates an example user interface that displays clinical information of interest according to an example embodiment herein.</li><li><figref idref="f0020">FIG. 16A</figref> illustrates an example 2D radiograph.</li><li><figref idref="f0020">FIG. 16B</figref> illustrates an example tomosynthesis image slice of the teeth of <figref idref="f0020">FIG. 16A</figref>, wherein the interproximal space is visible.</li><li><figref idref="f0021">FIG. 17A</figref> is a flowchart illustrating a process for generating a mask based on a two-dimensional orthogonal projection image and using the mask to guide a process for reducing image reconstruction artifacts in an intraoral tomosynthesis dataset according to an example embodiment herein.</li><li><figref idref="f0022">FIG. 17B</figref> is a flowchart illustrating a subprocess of <figref idref="f0021">FIG. 17A</figref>, namely features of step S1708 of <figref idref="f0021">FIG. 17A</figref>, according to an example embodiment herein.</li><li><figref idref="f0023">FIG. 17C</figref> is a flowchart illustrating a subprocess of <figref idref="f0021">FIG. 17A</figref>, namely features of step S1708 of <figref idref="f0021">FIG. 17A</figref>, according to another example embodiment herein.</li><li><figref idref="f0024">FIG. 18</figref> illustrates an example orthogonal projection image.</li><li><figref idref="f0025">FIG. 19</figref> illustrates a variance image corresponding to the orthogonal projection image of <figref idref="f0024">FIG. 18</figref>.</li><li><figref idref="f0026">FIG. 20</figref> illustrates a gradient image corresponding to the orthogonal projection image of <figref idref="f0024">FIG. 18</figref>.</li><li><figref idref="f0027">FIG. 21</figref> illustrates image statistics corresponding to the orthogonal projection image of <figref idref="f0024">FIG. 18</figref>.</li><li><figref idref="f0028">FIG. 22</figref> illustrates an example binary mask image corresponding to the orthogonal projection image of <figref idref="f0024">FIG. 18</figref>.</li><li><figref idref="f0029">FIG. 23</figref> illustrates a tomosynthesis image slice masked by the binary mask image of <figref idref="f0013">FIG. 8</figref>.</li><li><figref idref="f0030">FIG. 24</figref> illustrates a graph of an example intensity transformation function for compressing pixel values in isolated areas of a masked tomosynthesis image slice and a graph of an example intensity transformation function for linear mapping of pixel values in regions of the masked tomosynthesis image that contain anatomic information.</li><li><figref idref="f0031">FIG. 25</figref> illustrates a graph of an example intensity transformation function for sinusoidal mapping of pixel values in regions of the masked tomosynthesis image that contain anatomic information.</li><li><figref idref="f0032">FIG. 26</figref> illustrates an example reduced-artifact image slice.</li><li><figref idref="f0033">FIG. 27A</figref> is a flowchart illustrating a procedure for rendering a three-dimensional (3D) image from tomosynthesis slices according to an example embodiment herein.</li><li><figref idref="f0034">FIG. 27B</figref> is a flowchart illustrating another procedure for rendering a three-dimensional (3D) image from tomosynthesis slices according to an example embodiment herein.</li><li><figref idref="f0035">FIG. 27C</figref> is a flowchart illustrating one procedure for determining a three-dimensional outline trace of an object.</li><li><figref idref="f0035">FIG. 27D</figref> is a flowchart illustrating another procedure for determining a three-dimensional outline trace of an object.</li><li><figref idref="f0036">FIG. 28</figref> is a flowchart illustrating a procedure for pre-processing tomosynthesis slices according to an example embodiment herein.</li><li><figref idref="f0037">FIG. 29</figref> is a view for illustrating display of a 3D image according to an example embodiment herein.</li><li><figref idref="f0038">FIG. 30</figref> is a view for illustrating a 3D surface rendering based on a region-of-interest according to an example embodiment herein.</li><li><figref idref="f0039 f0040 f0041">FIGS. 31A to 31C</figref> are views for illustrating measurement of distances in a 3D image according to example embodiments herein.</li><li><figref idref="f0042">FIG. 32</figref> is a flow diagram for explaining measurement of distances in tomosynthesis images according to an example embodiment herein.</li><li><figref idref="f0043">FIGS. 33A-D</figref> are illustrations of a two-dimensional x-ray projection image, a tomosynthesis slice, a distance between the tomosynthesis slice and the x-ray detector, and the tomosynthesis slice within the image volume according to an example embodiment herein.</li><li><figref idref="f0044">FIG. 34</figref> is an image of a tomosynthesis slice according to an example embodiment herein.</li><li><figref idref="f0045">FIGS. 35A-D</figref> are illustrations of a two-dimensional x-ray projection image, another tomosynthesis slice, a distance between the tomosynthesis slice and the x-ray detector, and the other tomosynthesis slice within the image volume according to an example embodiment herein.</li><li><figref idref="f0046">FIG. 36</figref> is an illustration of the other tomosynthesis slice according to an example embodiment herein.</li><li><figref idref="f0047">FIGS. 37A-D</figref> are illustrations of a two-dimensional x-ray projection image, a tomosynthesis slice, a distance between the tomosynthesis slice and the x-ray detector, and a vector within the image volume according to an example embodiment herein.</li><li><figref idref="f0048">FIGS. 38A-B</figref> are illustrations of image volume with a vector therein from two different orientations according to an example embodiment herein.</li><li><figref idref="f0049">FIGS. 39A-C</figref> are illustrations of a two-dimensional x-ray projection image, a tomosynthesis slice, and a volumetric image of the image volume according to an example embodiment herein.</li><li><figref idref="f0050">FIGS. 40A-C</figref> are illustrations of a two-dimensional x-ray projection image, a tomosynthesis slice, and a volumetric image of the image volume with a vector disposed therein according to an example embodiment herein.</li><li><figref idref="f0051">FIGS. 41A-B</figref> are volumetric images of the image volume from different orientations with a vector disposed therein.</li><li><figref idref="f0052">FIG. 42A</figref> is an illustration of the human mandible.</li><li><figref idref="f0052">FIG. 42B</figref> is an illustration of region R1 in <figref idref="f0052">FIG. 42A</figref>.</li><li><figref idref="f0053">FIG. 43A</figref> is an illustration of the human maxilla from an occlusal viewpoint.</li><li><figref idref="f0053">FIG. 43B</figref> is an illustration of the human mandible from an occlusal viewpoint.</li><li><figref idref="f0054">FIG. 44</figref> is a flowchart illustrating an exemplary method of measuring the thickness of the lingual and buccal plates.</li><li><figref idref="f0055">FIG. 45</figref> is a perspective view of a portion of an exemplary tomosynthesis system <b>100.</b></li><li><figref idref="f0056">FIG. 46</figref> is an illustration of an x-ray source at different positions during an exemplary radiographic scan.</li><li><figref idref="f0057">FIG. 47</figref> is an illustration of an x-ray source positioned relative to a patient.</li><li><figref idref="f0058">FIG. 48</figref> is an illustration of an x-ray source positioned relative to a patient with the aid of an aiming device.</li><li><figref idref="f0059">FIG. 49</figref> is an illustration of an exemplary aiming device.</li><li><figref idref="f0060">FIG. 50</figref> is an illustration of an x-ray source at different positions during an exemplary radiographic scan.</li><li><figref idref="f0061">FIG. 51</figref> is an illustration of an x-ray source positioned relative to a patient.</li><li><figref idref="f0062">FIG. 52</figref> is an illustration of an x-ray source positioned relative to a patient with the aid of an aiming device.</li><li><figref idref="f0063">FIG. 53A</figref> is a perspective cross-sectional view of a human mandible.</li><li><figref idref="f0063">FIG. 53B</figref> is a cross-sectional view of the lingual and buccal plates shown in <figref idref="f0063">FIG. 53A</figref>.</li><li><figref idref="f0063">FIG. 53C</figref> is an exemplary two-dimensional image from a tomosynthesis stack.</li><li><figref idref="f0064">FIG. 54</figref> is a flow diagram illustrating a procedure for tracking motion of an object according to an example embodiment herein.</li><li><figref idref="f0065">FIG. 55</figref> is a flow diagram illustrating two methods for identifying objects in a projection image.</li><li><figref idref="f0066">FIG. 56A</figref> illustrates an example first projection image of teeth.</li><li><figref idref="f0066">FIG. 56B</figref> illustrates an example second projection image of teeth.</li><li><figref idref="f0067">FIG. 57A</figref> illustrates an example difference image determined from the first projection image of <figref idref="f0066">FIG. 56A</figref> and the second projection image of <figref idref="f0066">FIG. 56B</figref>.</li><li><figref idref="f0067">FIG. 57B</figref> illustrates an example binary image determined from the difference image of <figref idref="f0067">FIG. 57A</figref> and four objects identified thereon.</li><li><figref idref="f0068">FIG. 58</figref> illustrates an example projection image of teeth and a region of interest indicated thereon for each identified object.</li><li><figref idref="f0069">FIG. 59</figref> illustrates a correlation algorithm for determining the amount of shift for a region of interest.</li><li><figref idref="f0070">FIG. 60A</figref> is an illustration of a person positioned for an occlusal image of the mandible according to a conventional technique.</li><li><figref idref="f0070">FIG. 60B</figref> is a traditional two-dimensional x-ray image.</li><li><figref idref="f0070">FIG. 60C</figref> is an illustration of two x-rays propagating through matter.</li><li><figref idref="f0070">FIG. 60D</figref> is an illustration of lingual and buccal plates imaged according to a conventional technique.</li></ul>

0018Different ones of the Figures may have at least some reference numerals that are the same in order to identify the same components, although a detailed description of each such component may not be provided below with respect to each Figure.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

0019In accordance with example aspects described herein, methods, systems, apparatuses, and computer programs are provided for generating clinical information from a tomographic dataset, and more particularly, for identifying within an intraoral tomosynthesis dataset high-focus images containing features of interest. Tomosynthesis System

0020<figref idref="f0001">FIG. 1A</figref> illustrates a block diagram of an intraoral tomosynthesis system <b>100</b> for obtaining an intraoral tomosynthesis dataset, and which is constructed and operated in accordance with at least one example embodiment herein. The system <b>100</b> can be operated to obtain one or more x-ray images of an object <b>50</b> of interest, which may further include one or more sub-object(s) <b>52.</b> For example, object <b>50</b> may be a tooth (or teeth) and surrounding dentition of a patient, and sub-object(s) <b>52</b> may be root structures within the tooth.

0021The system <b>100</b> includes an x-ray detector <b>102</b> and an x-ray subsystem <b>116,</b> both of which, including subcomponents thereof, are electrically coupled to a computer system <b>106.</b> In one example, the x-ray subsystem <b>116</b> hangs from a ceiling or wall-mounted mechanical arm (not shown), so as to be freely positioned relative to an object <b>50.</b> The x-ray subsystem <b>116</b> further includes an x-ray source <b>104</b> mounted on a motorized stage <b>118</b> and an on-board motor controller <b>120.</b> The on-board motor controller <b>120</b> controls the motion of the motorized stage <b>118.</b>

0022The computer system <b>106</b> is electrically coupled to a display unit <b>108</b> and an input unit <b>114.</b> The display unit <b>108</b> can be an output and/or input user interface.

0023The x-ray detector <b>102</b> is positioned on one side of the object <b>50</b> and the receiving surface of the x-ray detector <b>102</b> extends in an x-y plane in a Cartesian coordinate system. The x-ray detector <b>102</b> can be a small intraoral x-ray sensor that includes, for example, a complementary metal-oxide semiconductor (CMOS) digital detector array of pixels, a charge-coupled device (CCD) digital detector array of pixels, or the like. In an example embodiment herein, the size of the x-ray detector <b>102</b> varies according to the type of patient to whom object <b>50</b> belongs, and more particularly, the x-ray detector <b>102</b> may be one of a standard size employed in the dental industry. Examples of the standard dental sizes include a "Size-2" detector, which is approximately 27×37 mm in size and is typically used on adult patients, a "Size-1" detector, which is approximately 21×31 mm in size and is typically used on patients that are smaller than Size-2 adult patients, and a "Size-0" detector, which is approximately 20×26 mm in size and is typically used on pediatric patients. In a further example embodiment herein, each pixel of the x-ray detector <b>102</b> has a pixel width of 15 µm, and correspondingly, the Size-2 detector has approximately 4 million pixels in a 1700×2400 pixel array, the Size-1 detector has approximately 2.7 million pixels in a 1300×2000 pixel array, and the Size-0 detector has approximately 1.9 million pixels in a 1200×1600 pixel array. The color resolution of the x-ray detector <b>102</b> may be, in one example embodiment herein, a 12-bit grayscale resolution, although this example is not limiting, and other example color resolutions may include an 8-bit grayscale resolution, a 14-bit grayscale resolution, and a 16-bit grayscale resolution.

0024The x-ray source <b>104</b> is positioned on an opposite side of the object <b>50</b> from the x-ray detector <b>102.</b> The x-ray source <b>104</b> emits x-rays <b>110</b> which pass through object <b>50</b> and are detected by the x-ray detector <b>102.</b> The x-ray source <b>104</b> is oriented so as to emit x-rays <b>110</b> towards the receiving surface of the x-ray detector <b>102</b> in at least a z-axis direction of the Cartesian coordinate system, where the z-axis is orthogonal to the x-y plane associated with the receiving surface of the x-ray detector <b>102.</b>

0025The x-ray source <b>104</b> can also emit x-rays <b>110</b> while positioned at each of multiple different locations within a scan angle <b>112,</b> where a 0° position in the scan angle <b>112</b> corresponds to the position for emitting x-rays <b>110</b> along the z-axis. In one example embodiment herein, the user initially positions the x-ray subsystem <b>116,</b> and hence, also the x-ray source <b>104,</b> to a predetermined starting position relative to the object <b>50.</b> The computer system <b>106</b> then controls the on-board motor controller <b>120</b> to move the x-ray source <b>104</b> via the motorized stage <b>118,</b> based on the known starting position, to step through each of the different locations within the scan angle <b>112.</b> The computer system <b>106</b> controls the x-ray source <b>104</b> to cause the source <b>104</b> to emit x-rays <b>110</b> at each of those locations.

0026The centroid of the x-rays <b>110</b> passes through a focal spot <b>122</b> at each of the different locations within the scan angle <b>112.</b> The focal spot <b>122</b> may be, for example, located close to the detector such that x-rays <b>110</b> emitted from the x-ray source <b>104</b> positioned at the outer limits of the scan angle <b>112</b> are aimed at and do not miss the x-ray detector <b>102.</b> In <figref idref="f0001">FIG. 1A</figref>, the 0° position is represented in x-ray source <b>104,</b> while reference numerals <b>104</b><i>a</i> and <b>104b</b> represent the same x-ray source <b>104</b> but in two other example positions within the scan angle <b>112.</b> The scan angle <b>112</b> can be, for example, ±20° from the 0° position, although this example is not limiting.

0027Additionally, the motion of x-ray source <b>104</b> along the scan angle <b>112</b> may form different scan paths, such as, for example, a linear scan <b>130</b> shown in <figref idref="f0002">FIG. 1B</figref>, a curved scan <b>132</b> shown in <figref idref="f0003">FIG. 1C</figref>, or a circular scan <b>134</b> shown in <figref idref="f0004">FIG. 1D</figref>. In the linear scan <b>130</b> (<figref idref="f0002">FIG. 1B</figref>), the x-ray source <b>104</b> moves linearly in an x-y plane while emitting x-rays <b>110</b> toward the focal spot <b>122,</b> forming a triangular sweep. In the curved scan <b>132</b> (<figref idref="f0003">FIG. 1C</figref>), the x-ray source <b>104</b> moves in an arc while emitting x-rays <b>110</b> toward the focal spot <b>122,</b> forming a fan beam sweep. In the circular scan <b>134</b> (<figref idref="f0004">FIG. 1D</figref>), the x-ray source <b>104</b> rotates around the z-axis while emitting x-rays <b>110</b> toward the focal spot <b>122,</b> forming a conical beam sweep. The scan positions also may be arranged in any particular one or more planes of the Cartesian coordinate system.

0028As emitted x-rays <b>110</b> pass through the object <b>50,</b> photons of x-rays <b>110</b> will be more highly attenuated by high density structures of the object <b>50,</b> such as calcium-rich teeth and bone, and less attenuated by soft tissues, such as gum and cheek. One or more of the attenuating structures can be sub-object(s) <b>52.</b> X-rays <b>110</b> passing through and attenuated by object <b>50,</b> are projected onto x-ray detector <b>102,</b> which converts the x-rays <b>110</b> into electrical signals and provides the electrical signals to computer system <b>106.</b> In one example embodiment, the x-ray detector <b>102</b> may be an indirect type of detector (e.g., a scintillator x-ray detector) that first converts x-rays <b>110</b> into an optical image and then converts the optical image into the electrical signals, and in another example embodiment, the x-ray detector <b>102</b> may be a direct type of detector (e.g., a semiconductor x-ray detector) that converts x-rays <b>110</b> directly into the electrical signals. The computer system <b>106</b> processes the electrical signals to form a two-dimensional projection image of the object <b>50.</b> In one example embodiment herein, the image size of the two-dimensional projection image corresponds to the dimensions and the number of pixels of the x-ray detector <b>102.</b>

0029The system <b>100</b> can collect a plurality of projection images, as described above, by first positioning the x-ray source <b>104</b> at different angles, including at least the 0° position, and emitting x-rays <b>110</b> at each of those different angles through object <b>50</b> towards x-ray detector <b>102.</b> For example, the plurality of projection images may include a total of fifty-one projections: one orthogonal projection image, obtained when the x-ray source is at the 0° position, and fifty projection images, each obtained when the x-ray source <b>104</b> is positioned at different angles within a range of ±20° from the z-axis (corresponding to the scan angle <b>112</b>). In other example embodiments, the number of projection images may range from twenty-five to seventy. Because the orthogonal projection image is obtained when the x-ray source is at the 0° position, the orthogonal projection image has the same appearance as a conventional x-ray image. That is, the two-dimensional orthogonal projection image has no depth perception, and one or more sub-object(s) <b>52</b> within object <b>50</b> may appear overlaid one on top of another in the orthogonal projection image, as represented in <figref idref="f0005">FIG. 1E</figref>, for example. On the other hand, sub-object(s) <b>52</b> at different depths of the z-axis within object <b>50</b> undergo varying degrees of parallax when imaged from different angles along the scan angle <b>112,</b> as represented in <figref idref="f0006">FIG. 1F</figref>, for example.

0030The computer system <b>106</b> processes the plurality of projection images to reconstruct a series of two-dimensional tomosynthesis image slices, also known as a tomosynthesis stack of images, in a manner to be described below. Each image slice is parallel to the plane in which the receiving surface of the x-ray detector <b>102</b> extends and at different depths of the z-axis.

0031The computer system <b>106</b> further processes the tomosynthesis image slices in a manner to be described below, to generate clinically relevant information related to object <b>50</b> (e.g., a patient's dental anatomy), and in a further example embodiment herein, related to sub-object(s) <b>52.</b> The extracted information may include the identification, within the tomosynthesis stack of images, of high-focus images that contain features of interest therein. In one example embodiment herein, the computer system <b>106</b> obtains input from a user via input unit <b>114</b> and/or display unit <b>108</b> to guide the further processing of the tomosynthesis slices.

0032The orthogonal projection image, one or more image slices of the tomosynthesis stack, and the extracted information are provided by the computer system <b>106</b> for display to the user on the display unit <b>108.</b>

0033Compared to a dental CBCT system, the intraoral tomosynthesis imaging system <b>100</b> carries a lower cost of ownership, can acquire images faster and with higher resolution (e.g., a per pixel resolution of approximately 20 µm, compared to a per pixel resolution of 100-500 µm with CBCT), and exposes patients to a lower x-ray dose (e.g. approximately an order of magnitude lower in some cases, owing in part to a smaller field of view, a smaller scan angle, and the need to only penetrate the anatomy between the x-ray source <b>104</b> and the x-ray detector <b>102,</b> rather than the complete jaw). Additionally, in some example embodiments herein, the intraoral tomosynthesis system <b>100</b> can resemble a conventional x-ray radiography system, and can use the same or substantially similar equipment, such as, for example, the ceiling- or wall-mounted mechanical arm for positioning the x-ray source <b>104,</b> a similarly-sized x-ray source <b>104,</b> and the intraoral x-ray detector <b>102.</b> Accordingly, operation of the intraoral tomosynthesis system <b>100</b> is more familiar and less complex to a clinician, compared to dental CBCT, and also can be used chair-side.

Computer System for Tomosynthesis Imaging

0034Having described a system <b>100</b> for acquiring a tomosynthesis dataset and for generating clinically relevant information from a tomosynthesis dataset, including the identification of high-focus images containing features of interest, reference will now be made to <figref idref="f0007">FIG. 2A</figref>, which shows a block diagram of a computer system <b>200</b> that may be employed in accordance with at least some of the example embodiments herein. Although various embodiments are described herein in terms of this exemplary computer system <b>200,</b> after reading this description, it will become apparent to a person skilled in the relevant art(s) how to implement the invention using other computer systems and/or architectures.

0035<figref idref="f0007">FIG. 2A</figref> illustrates a block diagram of the computer system <b>200.</b> In one example embodiment herein, at least some components of the computer system <b>200</b> (such as all those components, or all besides component <b>228</b>) can form or be included in the computer system <b>106</b> shown in <figref idref="f0001">FIG. 1A</figref>. The computer system <b>200</b> includes at least one computer processor <b>222</b> (also referred to as a "controller"). The computer processor <b>222</b> may include, for example, a central processing unit, a multiple processing unit, an application-specific integrated circuit ("ASIC"), a field programmable gate array ("FPGA"), or the like. The processor <b>222</b> is connected to a communication infrastructure <b>224</b> (e.g., a communications bus, a cross-over bar device, or a network).

0036The computer system <b>200</b> also includes a display interface (or other output interface) <b>226</b> that forwards video graphics, text, and other data from the communication infrastructure <b>224</b> (or from a frame buffer (not shown)) for display on a display unit <b>228</b> (which, in one example embodiment, can form or be included in the display unit <b>108</b>). For example, the display interface <b>226</b> can include a video card with a graphics processing unit.

0037The computer system <b>200</b> also includes an input unit <b>230</b> that can be used by a user of the computer system <b>200</b> to send information to the computer processor <b>222.</b> In one example embodiment herein, the input unit <b>230</b> can form or be included in the input unit <b>114.</b> For example, the input unit <b>230</b> can include a keyboard device and/or a mouse device or other input device. In one example, the display unit <b>228,</b> the input unit <b>230,</b> and the computer processor <b>222</b> can collectively form a user interface.

0038In an example embodiment that includes a touch screen, for example, the input unit <b>230</b> and the display unit <b>228</b> can be combined, or represent a same user interface. In such an embodiment, a user touching the display unit <b>228</b> can cause corresponding signals to be sent from the display unit <b>228</b> to the display interface <b>226,</b> which can forward those signals to a processor such as processor <b>222,</b> for example.

0039In addition, the computer system <b>200</b> includes a main memory <b>232,</b> which preferably is a random access memory ("RAM"), and also may include a secondary memory <b>234.</b> The secondary memory <b>234</b> can include, for example, a hard disk drive <b>236</b> and/or a removable-storage drive <b>238</b> (e.g., a floppy disk drive, a magnetic tape drive, an optical disk drive, a flash memory drive, and the like). The removable-storage drive <b>238</b> reads from and/or writes to a removable storage unit <b>240</b> in a well-known manner. The removable storage unit <b>240</b> may be, for example, a floppy disk, a magnetic tape, an optical disk, a flash memory device, and the like, which is written to and read from by the removable-storage drive <b>238.</b> The removable storage unit <b>240</b> can include a non-transitory computer-readable storage medium storing computer-executable software instructions and/or data.

0040In alternative embodiments, the secondary memory <b>234</b> can include other computer-readable media storing computer-executable programs or other instructions to be loaded into the computer system <b>200.</b> Such devices can include a removable storage unit <b>244</b> and an interface <b>242</b> (e.g., a program cartridge and a cartridge interface similar to those used with video game systems); a removable memory chip (e.g., an erasable programmable read-only memory ("EPROM") or a programmable read-only memory ("PROM")) and an associated memory socket; and other removable storage units <b>244</b> and interfaces <b>242</b> that allow software and data to be transferred from the removable storage unit <b>244</b> to other parts of the computer system <b>200.</b>

0041The computer system <b>200</b> also can include a communications interface <b>246</b> that enables software and data to be transferred between the computer system <b>200</b> and external devices. Examples of the communications interface <b>246</b> include a modem, a network interface (e.g., an Ethernet card or an IEEE 802.11 wireless LAN interface), a communications port (e.g., a Universal Serial Bus ("USB") port or a FireWire^® port), a Personal Computer Memory Card International Association ("PCMCIA") interface, and the like. Software and data transferred via the communications interface <b>246</b> can be in the form of signals, which can be electronic, electromagnetic, optical or another type of signal that is capable of being transmitted and/or received by the communications interface <b>246.</b> Signals are provided to the communications interface <b>246</b> via a communications path <b>248</b> (e.g., a channel). The communications path <b>248</b> carries signals and can be implemented using wire or cable, fiber optics, a telephone line, a cellular link, a radio-frequency ("RF") link, or the like. The communications interface <b>246</b> may be used to transfer software or data or other information between the computer system <b>200</b> and a remote server or cloud-based storage (not shown).

0042One or more computer programs (also referred to as computer control logic) are stored in the main memory <b>232</b> and/or the secondary memory <b>234.</b> The computer programs also can be received via the communications interface <b>246.</b> The computer programs include computer-executable instructions which, when executed by the computer processor <b>222,</b> cause the computer system <b>200</b> to perform the procedures as described herein (and shown in figures), for example. Accordingly, the computer programs can control the computer system <b>106</b> and other components (e.g., the x-ray detector <b>102</b> and the x-ray source <b>104</b>) of the intraoral tomosynthesis system <b>100.</b>

0043In one example embodiment herein, the software can be stored in a non-transitory computer-readable storage medium and loaded into the main memory <b>232</b> and/or the secondary memory <b>234</b> of the computer system <b>200</b> using the removable-storage drive <b>238,</b> the hard disk drive <b>236,</b> and/or the communications interface <b>246.</b> Control logic (software), when executed by the processor <b>222,</b> causes the computer system <b>200,</b> and more generally the intraoral tomosynthesis system <b>100,</b> to perform the procedures described herein.

0044In another example embodiment hardware components such as ASICs, FPGAs, and the like, can be used to carry out the functionality described herein. Implementation of such a hardware arrangement so as to perform the functions described herein will be apparent to persons skilled in the relevant art(s) in view of this description.

0045Having provided a general description of the tomosynthesis system <b>100,</b> techniques for processing data from the tomosynthesis system <b>100</b> (or as the case may be a CT or CBCT machine as well) will be described below. As one of ordinary skill will appreciate, description corresponding to one technique may be applicable to another technique described herein.

Generating Clinical Information from a Dataset

0046Generally, for x-ray images to have value and utility in clinical diagnosis and treatment, they should have high image fidelity and quality (as measured by resolution, brightness, contrast, signal-to-noise ratio, and the like, although these example metrics are not limiting) so that anatomies of interest can be clearly identified, analyzed (e.g., analysis of shape, composition, disease progression, etc.), and distinguished from other surrounding anatomies.

0047In addition to providing tomosynthesis image slices with good image fidelity and quality (although such is not necessary), an intraoral tomosynthesis system <b>100</b> according to example aspects herein augments the tomosynthesis image slices by automatically or semi-automatically generating clinical information of interest about the imaged object <b>50</b> and presenting the same to the clinician user. In an example embodiment herein, the clinical information of interest relates to anatomical features (such as sub-object(s) <b>52</b>) located at a depth within the object <b>50,</b> and such anatomical features may not be readily apparent in the tomosynthesis image slices under visual inspection by the clinician user and also may not be visible in a conventional 2D radiograph due to overlapping features from other depths.

0048The intraoral tomosynthesis system <b>100</b> will now be further described in conjunction with <figref idref="f0008">FIG. 2B</figref>, which shows a flow diagram of a process for generating clinical information of interest according to an example embodiment herein.

0049The process of <figref idref="f0008">FIG. 2B</figref> starts at Step S<b>201,</b> and in Step S<b>202,</b> the tomosynthesis system <b>100</b> acquires a plurality of projection images of the object <b>50</b> over a scan angle <b>112.</b>

0050In Step S<b>204,</b> the computer system <b>106</b> processes the plurality of projection images to reconstruct a series of two-dimensional tomosynthesis image slices (also known as a tomosynthesis stack), each image slice representing a cross-section of the object <b>50</b> that is parallel to the x-ray detector <b>102</b> and each slice image also being positioned at a different, respective, location along the z-axis (i.e., in a depth of the object <b>50</b>) than other image slices. (The reconstruction of the tomosynthesis stack in Step S<b>204</b> can be substantially the same process as that of Step S<b>304</b> of <figref idref="f0009">FIG. 3</figref> described in greater detail herein below.)

0051In Step S<b>206,</b> the computer system <b>106</b> receives, via input unit <b>114</b> and/or display unit <b>108,</b> a guidance from a clinician user indicating a clinical aspect of interest. In an example embodiment herein, the received guidance may be a user selection from among a predetermined list of tools presented by the computer system <b>106.</b>

0052The guidance received in Step S<b>206</b> may be, for example, and without limitation, a selection of at least one region of interest on at least one of the projection images or the tomosynthesis image slices, at least one anatomy of interest (e.g., mental foramen, nerve canal, sinus floor, sinus cavity, nasal cavity, periodontal ligament, lamina dura, or other dental anatomies), a type of dental procedure (e.g., an endodontic procedure, a periodontic procedure, an implantation, caries detection, crack detection, and the like), a measurement inquiry (e.g., a distance measurement, a volumetric measurement, a density measurement, and the like), or any combination thereof

0053In Step S<b>208,</b> the computer system <b>106</b> processes the tomosynthesis stack to generate information that is relevant to the clinical aspect of interest indicated by the guidance received in Step S<b>206.</b> In an example embodiment herein, the computer system <b>106</b> performs a processing in Step S<b>208</b> that is predetermined to correspond to the received guidance.

0054Non-limiting examples of tomosynthesis stack processing that can be performed in Step S<b>208</b> (and the information generated thereby) for a particular received guidance are as follows.

0055In an example embodiment herein where the received guidance is a selection of at least one region of interest on at least one of the projection images or the tomosynthesis image slices, the computer system <b>106</b> processes the tomosynthesis stack according to a process described further herein below with reference to <figref idref="f0009">FIG. 3</figref>.

0056Where the received guidance is at least one anatomy of interest (e.g., mental foramen, nerve canal, sinus floor, sinus cavity, nasal cavity, periodontal ligament, lamina dura, or other dental anatomies), the computer system <b>106</b> processes the tomosynthesis stack to identify the anatomy of interest (e.g., by way of image segmentation). One or more image segmentation techniques may be used to identify the anatomy of interest including, for example, a Hough transformation, a gradient segmentation technique, and a minimal path (geodesic) technique, which are discussed in further detail below. The computer system <b>106</b> generates, as generated information, a display image that indicates the anatomy of interest (e.g., by highlighting, outlining, or the like). In one example embodiment herein, the display can be the tomosynthesis image slices with the identified anatomy indicated thereon or a 3D rendering of the identified anatomy. For example, <figref idref="f0015">FIG. 11</figref> illustrates a tomosynthesis image slice with a nerve canal <b>1102</b> outlined thereon, and <figref idref="f0016">FIG. 12</figref> illustrates a tomosynthesis image slice with a sinus cavity <b>1202</b> outlined thereon. As another example, <figref idref="f0017">FIG. 13A</figref> illustrates a 2D radiograph of a patient's anatomy, wherein a nasal cavity is less clearly defined, but, by virtue of performing Step S<b>208</b> on a tomosynthesis dataset acquired from the same anatomy, the computer system <b>106</b> identifies the nasal cavity and indicates at least the nasal cavity walls <b>1302, 1304, 1306,</b> and <b>1308</b> on the tomosynthesis image slices shown in <figref idref="f0017">FIGS. 13B and 13C</figref>.

0057If the received guidance is a type of dental procedure (e.g., an endodontic procedure, a periodontic procedure, an implantation, caries detection, crack detection, and the like), the computer system <b>106</b> generates information specific to the dental procedure.

0058For example, for a guidance indicating an endodontic root canal procedure, the computer system <b>106</b> processes the tomosynthesis dataset to identify root canals and generates a display of the identified root canals as the generated information (as discussed below). For example, the generated information can be the tomosynthesis image slices with the root canals highlighted and/or a 3D rendering of the root canals. In an additional example embodiment herein, the computer system <b>106</b> can generate spatial information related to the shape of the root canal, such as, for example, its location, curvature, and length.

0059For a received guidance indicating an implantation, the computer system <b>106,</b> in an example embodiment herein, processes the tomosynthesis stack and generates, as the generated information, locations of anatomical landmarks of interest for an implant procedure, such as, for example, a location of the nerve canal, a location of the sinus floor, a location of the gingival margin, and a location of the buccal plate, through image segmentation. The computer system <b>106</b> can also generate, as the generated information, a 3D rendering of the jaw with the teeth virtually extracted.

0060For a received guidance indicating caries detection, the computer system <b>106,</b> in an example embodiment herein, processes the tomosynthesis stack to detect caries and generates, as the generated information, the locations of carious lesion(s). In one embodiment, the guidance may include information that the computer system <b>106</b> uses to evaluate segmented regions and identify one or more of the regions as carious regions. Such information may include, for example, expected region size and attenuation amounts for a carious region. The locations of carious lesion(s) can be in the form of the tomosynthesis image slices with the carious region(s) highlighted thereon or a 3D rendering of the affected tooth of teeth with the carious volume(s) highlighted thereon.

0061For a received guidance indicating crack detection, the computer system <b>106,</b> in an example embodiment herein, processes the tomosynthesis stack to detect cracks and generates, as the generated information, the location of any cracks in the imaged tooth or teeth. In some example embodiments herein, the location of a crack can be in the form of the tomosynthesis image slices with the crack indicated thereon or a 3D rendering of the affected tooth of teeth with the crack indicated thereon. For example, the computer system <b>106</b> can process a tomosynthesis dataset to identify cracks in the imaged teeth (using image segmentation), and then generate the tomosynthesis image slices shown in <figref idref="f0018">FIGS. 14B and 14C</figref> with the identified cracks <b>1402</b> and <b>1404</b> indicated thereon, respectively.

0062In an example embodiment herein where the received guidance is a measurement inquiry (e.g., a distance measurement, an 2D area or 3D volumetric measurement, a density measurement, and the like), the computer system <b>106</b> processes the tomosynthesis stack to calculate the requested measurement as the generated information. For example, the computer system <b>100</b> can calculate, as the generated information, a distance between at least two user-selected points in the tomosynthesis dataset, a distance between two or more anatomies identified in the manner described above, an area or volume of an identified anatomy or of a user-selected region of the tomosynthesis dataset, or a density of an identified anatomy or of a region of the tomosynthesis dataset.

0063In Step S<b>210,</b> the computer system <b>106</b> presents the information generated in Step S<b>208</b> to the user on the display unit <b>108.</b> In an example embodiment herein, the computer system <b>106</b> can present the information generated in Step S<b>208</b> by way of a user interface displayed on display unit <b>108.</b>

0064<figref idref="f0019">FIG. 15</figref> illustrates a particular example of a user interface for presenting, in accordance with Step S<b>210,</b> information generated in Step S<b>208</b> in response to a guidance received in Step S<b>206</b> to locate the mental foramen <b>1502.</b> In the particular example shown in <figref idref="f0019">FIG. 15</figref>, the computer system <b>106</b> displays a tomosynthesis image slice <b>1504</b> with the location of the mental foramen <b>1502</b> indicated thereon, a 3D rendering of a tooth <b>1506</b> with the location of the mental foramen <b>1502</b> indicated in 3D space in relation to the 3D-rendered tooth <b>1506,</b> and a distance measurement <b>1508</b> from the apex of the 3D-rendered tooth <b>1506</b> to the mental foramen <b>1502.</b>

0065The process of <figref idref="f0008">FIG. 2B</figref> ends at Step S<b>212.</b>

0066As can be appreciated in view of the foregoing, by virtue of the processing being performed on a tomosynthesis stack, which includes 3D information about the object <b>50</b> as explained above, the generated information also provides to the clinician user a depth information and depth context about the object <b>50</b> that may not be readily apparent in the tomosynthesis image slices under visual inspection by the clinician user and also may not be visible in a conventional 2D radiograph due to overlapping features from other depths.

0067As one particular example of useful depth information provided to a user, the tomosynthesis system <b>100</b> performing the process of <figref idref="f0008">FIG. 2B</figref> can automatically detect interproximal caries between teeth, because an interproximal space (e.g., space <b>1602</b> on <figref idref="f0020">FIG. 16B</figref>) between teeth is visible in at least one of the tomosynthesis image slices but would be obscured by overlapping anatomies in a conventional 2D radiograph of the same region (e.g., <figref idref="f0020">FIG. 16A</figref>). As another particular example of a depth information, the tomosynthesis system <b>100</b> performing the process of <figref idref="f0008">FIG. 2B</figref> can automatically detect dental cracks (e.g., cracks <b>1402</b> and <b>1404</b> on <figref idref="f0018">FIGS. 14B and 14C</figref>, respectively) in individual ones of the tomosynthesis image slices, which also may be obscured by overlapping anatomies in a conventional 2D radiograph (e.g., <figref idref="f0018">FIG. 14A</figref>).

0068Additionally, by virtue of using the computer system <b>106</b> to perform at least part of the process shown in <figref idref="f0008">FIG. 2B</figref> and described above, the tomosynthesis system <b>100</b> can be controlled to acquire images of lower fidelity and lower quality, thus potentially lowering the x-ray exposure to the patient and reducing image acquisition time, even while generating and presenting clinical information of high value and utility.

Identifying High-Focus Images within a Dataset

0069The intraoral tomosynthesis system <b>100</b> will now be further described in conjunction with <figref idref="f0009">FIG. 3</figref>, which shows a flow diagram of a process according to an example embodiment herein for identifying high-focus images within a tomosynthesis dataset. Prior to starting the process, the x-ray detector <b>102</b> and x-ray source <b>104</b> are aligned manually by a user to a starting position, as described above, in one example embodiment herein.

0070The process of <figref idref="f0009">FIG. 3</figref> starts at Step S<b>301,</b> and, in Step S<b>302,</b> the intraoral tomosynthesis system <b>100</b> acquires a plurality of projection images of object <b>50</b> over a scan angle <b>112</b> (which may be predetermined), including the orthogonal projection image, in the manner described above. For example, the x-ray source <b>104</b> is moved by the motorized stage <b>118</b> and control circuitry <b>120</b> to different positions within the scan angle <b>112,</b> and the computer system <b>106</b> controls the x-ray source <b>104</b> to emit x-rays <b>110</b> at each position. In one example embodiment herein, x-ray source <b>104</b> is scanned, by pivoting at a point along the z-axis, from -20° from the z-axis to +20° from the z-axis in evenly distributed increments of 0.8° to provide <b>51</b> scan angles, including the 0° position, although this example is not limiting. The x-rays <b>110</b> then pass through and are attenuated by the object <b>50</b> before being projected onto the x-ray detector <b>102.</b> The x-ray detector <b>102</b> converts the x-rays <b>110</b> into electrical signals (either directly or indirectly, as described above) and provides the electrical signals to the computer system <b>106.</b> The computer system <b>106</b> processes the electrical signals collected at each scan angle position to acquire the plurality of projection images, each image comprising an array of pixels. The image acquired with the x-ray source <b>104</b> at the 0° position is also referred to herein as an orthogonal projection image.

0071In one example embodiment herein, the color depth of each pixel value of the projection images may be 12-bit grayscale, and the dimensions of the projection images correspond to the standard dental size of the x-ray detector <b>102,</b> as described above. For example, a Size-2 detector may produce projection images that are approximately 1700×2400 pixels in size, a Size-1 detector may produce projection images that are approximately 1300×2000 pixels in size, and a Size-0 detector may produce projection images that are approximately 1200×1600 pixels in size.

0072In Step S<b>304,</b> the computer system <b>106</b> processes the plurality of projection images acquired in Step S<b>302</b> using a reconstruction technique in order to reconstruct a series of two-dimensional tomosynthesis image slices and may also perform deblurring and other image enhancements, as will be described further herein. Each reconstructed image slice is a tomographic section of object <b>50</b> comprising an array of pixels, that is, each image slice represents a cross-section of object <b>50</b> that is parallel to the x-y plane in which the receiving surface of the x-ray detector <b>102</b> extends, has a slice thickness along the z-axis, and is positioned at a different, respective location along the z-axis than other image slices. The slice thickness is a function of the reconstruction technique and aspects of the geometry of the system <b>100,</b> including, primarily, the scan angle <b>112.</b> For example, each image slice may have a slice thickness of 0.5 mm by virtue of the geometry of the system <b>100</b> and the reconstruction technique. The desired location of each reconstructed image slice along the z-axis is provided as an input to the reconstruction performed in Step S<b>304</b> either as a pre-programmed parameter in computer system <b>106</b> or by user input via input unit <b>114</b> and/or display unit <b>108.</b> By example only, the computer system <b>106</b> can be instructed to reconstruct, from the plurality of projection images, a first image slice that is one millimeter (1 mm) away from the surface of x-ray detector <b>102</b> along the z-axis, a last image slice being at fifteen millimeters (15 mm) away from the surface of the x-ray detector <b>102,</b> and image slices between the first image slice and the last image slice at regular increments along the z-axis of two-hundred micrometers (200 µm), for a total of seventy-one image slices.

0073Reconstruction of the tomosynthesis image slices in Step S<b>304</b> may be performed in accordance with any existing or later developed reconstruction technique. For example, a shift-and-add method, filtered backprojection, matrix inversion tomosynthesis, generalized filtered backprojection, SIRT (simultaneous iterative reconstruction technique), or algebraic technique, among others, may be used. In one example embodiment herein, reconstruction of the tomosynthesis image slices in Step S<b>304</b> utilizes a shift-and-add technique. The shift-and-add technique utilizes information about the depth of sub-object(s) <b>52</b> along the z-axis that is reflected in the parallax captured by the plurality of projection images, as described above. According to this example embodiment, an image slice is reconstructed by first spatially shifting each projection image by an amount that is geometrically related to the distance between the image slice and the focal spot <b>122</b> along the z-axis. The shifted projection images are then averaged together to result in the image slice, where all sub-objects <b>52</b> in the plane of the image slice are in focus and sub-objects <b>52</b> outside of that plane are out of focus and blurry. This shift-and-add process is repeated for each image slice to be reconstructed. In the case of the image slice corresponding to the x-y plane that includes the focal spot <b>122,</b> the projection images are averaged together without first shifting because sub-objects <b>52</b> are already in focus for that plane.

0074The foregoing describes a basic shift-and-add reconstruction technique. In one example embodiment herein, a deblurring technique that substantially reduces or removes blurry, out-of-plane sub-objects from an image slice can be performed in conjunction with the reconstruction technique (whether shift-and-add or another technique). Examples of deblurring techniques that can be employed include, for example, spatial frequency filtering, ectomography, filtered backprojection, selective plane removal, iterative restoration, and matrix inversion tomosynthesis, each of which may be used in Step S<b>304</b> to deblur images reconstructed by the shift-and-add reconstruction technique (or another reconstruction technique, if employed).

0075In another example embodiment herein, Step S<b>304</b> also can include the computer system <b>106</b> performing further automated image enhancements such as, for example, image sharpening, brightness optimization, and/or contrast optimization, on each reconstructed (and deblurred, where deblurring is performed) image slice in a known manner.

0076Additionally, in another example embodiment herein, the dimensions, position, and orientation of each image slice reconstructed in Step S<b>304</b> are the same as the corresponding characteristics of the orthogonal projection image. Thus, when tomosynthesis image slices (or portions thereof) and the orthogonal projection image are overlaid over one another, corresponding anatomical features appearing in the images will be overlapped and aligned without scaling, rotation, or other transformation of the images.

0077In Step S<b>306,</b> the computer system <b>106</b> assembles the tomosynthesis image slices into an ordered stack of two-dimensional tomosynthesis images slices. Each image slice is assembled into the stack according to its corresponding location in object <b>50</b> along the z-axis, such that the image slices in the stack are ordered along the z-axis in the order of such locations along that axis. Each image slice is associated with an image number representing the position of that image in the ordered stack. For example, in a stack of sixty tomosynthesis image slices assembled from sixty tomosynthesis image slices, image number one can be the image slice closest to the x-ray detector <b>102</b> and image number sixty can be the image slice farthest from the x-ray detector <b>102.</b> In one example embodiment herein, images of the plurality of projection images and image slices of the tomosynthesis stack have the same dimensional resolution and color depth characteristics.

0078After Step S<b>306,</b> control passes to Step S<b>310,</b> which will be described below. Before describing that step, Step S<b>308</b> will first be described. Like Step S<b>304,</b> Step S<b>308</b> is performed after Step S<b>302</b> is performed.

0079In Step S<b>308,</b> the orthogonal projection image is extracted from the plurality of projection images acquired in Step S<b>302.</b> Because, as described above, the orthogonal projection image is defined as the projection image captured while the x-ray source <b>104</b> is in the 0° scan angle position, no reconstruction is necessary to extract that image. In one example embodiment herein, the orthogonal projection image is extracted and stored in the main memory <b>232,</b> although it may be stored instead in the secondary memory <b>234,</b> and can be retrieved therefrom for display in Step S<b>310</b> and/or Step S<b>322.</b> In another example embodiment herein, the extracted orthogonal projection image may undergo automated image enhancements (performed by computer system <b>106</b>) such as, for example, image sharpening, brightness optimization, and/or contrast optimization, in a known manner.

0080In Step S<b>310,</b> the stack of tomosynthesis image slices assembled in Step S<b>306</b> and the orthogonal projection image extracted in Step S<b>308</b> are displayed on the display unit <b>108.</b> In one example embodiment herein, the displaying can be performed as to show the entire stack, or one or more selected image slices of the stack, using display unit <b>108,</b> and interactive controls (e.g. via display unit <b>108</b> and/or input device <b>114</b>) enable a user to select between those two options, and to select one or more image slices for display, and also to select one or more particular regions of interest in the image(s) for display (whether in zoom or non-zoom, or reduced fashion). In a further example embodiment, as described below, stack controls <b>1016</b> illustrated in <figref idref="f0014">FIG. 10</figref> are provided and can include a scroll bar, which enables the user to manually select which image slice is displayed on the display unit <b>108,</b> and/or can include selectable control items, such as play, pause, skip forward, and skip backward, (not shown) to enable the user to control automatic display of the tomosynthesis stack, as a cine loop for example, on the display unit <b>108.</b>

0081In Step S<b>312,</b> the computer system <b>106</b> receives, via input unit <b>114</b> and/or display unit <b>108,</b> an indication of a region of interest from a user. In one example embodiment herein, the user indicates a region of interest on the orthogonal projection image displayed on the display unit <b>108</b> in Step S<b>310.</b> In an alternative example embodiment herein, the user indicates a region of interest on a tomosynthesis image slice displayed on the display unit <b>108</b> in Step S<b>310.</b>

0082Additionally, the region of interest may be a rectangular marquee (or any other outlining tool, including but not limited to a hand-drawn outline, a marquee of a predetermined shape, and the like) drawn on the orthogonal projection image or tomosynthesis image slice displayed on the display unit <b>108</b> in Step S<b>310.</b> For example, <figref idref="f0010">FIG. 4</figref> illustrates an example of a rectangular region of interest <b>402</b> drawn over an example orthogonal projection image, although this example is not limiting.

0083In Step S<b>314,</b> the computer system <b>106</b> applies a focus function to determine the degree to which the region of interest of each image slice in the tomosynthesis stack is in focus and assigns a focus factor to each image slice based on the results of the focus function. In one example embodiment herein, prior to applying the focus function, the computer system <b>106</b> pre-processes image slices in the tomosynthesis stack to reduce image artifacts, such as ringing, motion blur, hot-pixels, and x-ray generated noise. In a further example embodiment herein, the image pre-processing includes applying a Gaussian blur filter to each image slice in a known manner.

0084After pre-processing image slices, if performed, the computer system <b>106</b> applies the focus function to each image slice in the tomosynthesis stack. For example, first, the computer system <b>106</b> extracts a region of interest image, which is a portion of the tomosynthesis slice image corresponding to the region of interest received in Step S<b>312.</b> Then, the region of interest image is padded on all sides to avoid or substantially minimize possible creation (if any) of image processing artifacts in a border region during subsequent processing in Step S<b>312,</b> including the deriving of a variance image as described below. The pixel values of the padding may be, for example, a constant value (e.g., zero), an extension of the border pixels of the region of interest image, or a mirror image of the border pixels of the region of interest image. After the region of interest image has been padded, a variance image is derived by iterating a variance kernel operator, for example, a 5×5 pixel matrix, through each pixel coordinate of the region of interest image. At each iterative pixel coordinate, the statistical variance of pixel values of the region of interest image within the variance kernel operator is calculated, and the result is assigned to a corresponding pixel coordinate in the variance image. Then, the variance image is cropped to the same size as that of the unpadded region of interest image. Lastly, the focus factor is calculated as the statistical mean of the pixel values in the cropped variance image. Accordingly, a high focus factor corresponds to a high mean variance within the region of interest image. The focus factor is assigned to the image slice, and correspondingly, the focus factor is associated with the image number of the image slice to which it is assigned. The foregoing process is applied to each slice, for example, by serial iteration and/or in parallel, to assign a focus factor to each image slice.

0085In the preceding example embodiment, performing the focus function on the region of interest portion of each image slice instead of on the full view of each image slice facilitates the process of <figref idref="f0009">FIG. 3</figref> in identifying the z-axis location of images slices that are in high focus within the region of interest relative to other images slices. The focus function can be performed on the full view of image slices; however, focusing techniques performed on the full view may, in at least some cases, function less effectively to identify the location of high-focus images, because the full view of most, if not all, image slices contains both in-focus and out-of-focus regions.

0086In Step S<b>316,</b> the computer system <b>106</b> creates a focus profile from a series of the focus factors assigned in Step S<b>314,</b> where the focus factors are ordered in the focus profile according to their corresponding image numbers. <figref idref="f0011">FIG. 5</figref> illustrates an example focus profile <b>502</b> (shown as a solid line) within the region of interest <b>402</b> of <figref idref="f0010">FIG. 4</figref>. The focus profile <b>502</b> of <figref idref="f0011">FIG. 5</figref> is shown as a graph plotting the focus factor on the left side y-axis for each image slice number on the x-axis.

0087In Step S<b>318,</b> the computer system <b>106</b> searches for a local extremum in the focus profile (i.e., a local maximum or a local minimum), using known techniques. A focus profile may have more than one local extremum. For example, <figref idref="f0011">FIG. 5</figref> illustrates, in addition to the example of the focus profile <b>502,</b> local maxima <b>504</b> and <b>506</b> identified as a result of Step S<b>318</b> being performed.

0088In one example embodiment herein, the computer system <b>106</b> compares each focus factor to its neighbors iteratively, wherein neighbors are defined as the focus factors within a predetermined range of image numbers of the focus factor being evaluated during an iteration. If the focus factor being evaluated is greater than the individual focus factors of all of its neighbors, the focus factor being evaluated is designated a local maximum; otherwise, it is not designated a local maximum.

0089In another example embodiment herein, the computer system <b>106</b> performs a first derivative test to search for the local maximum of the focus profile. A first derivative of the focus profile is calculated from the focus profile (e.g., calculating a difference value at each image number of the focus profile by subtracting the focus factor at one image number from the focus factor of the next greater image number), and then the local maximum is identified as corresponding to the image number where the first derivative of the focus profile crosses zero from positive to negative. For example, <figref idref="f0011">FIG. 5</figref> illustrates a first derivative <b>508</b> (shown as a dot-dash line, with magnitude of the first derivative on the right side y-axis) corresponding to the focus profile <b>502.</b> Local maxima <b>504</b> and <b>506</b> are identified as corresponding to first derivative zero crossings <b>510</b> and <b>512,</b> respectively, as a result of performing Step <b>S318</b> according to an example of the present embodiment.

0090In a further example embodiment herein, the focus profile is filtered, that is, smoothed, by a moving average before searching for a local maximum in the focus profile. <figref idref="f0011">FIG. 5</figref> illustrates a filtered (smoothed) focus profile <b>514</b> (shown as a dotted line) corresponding to the focus profile <b>502.</b> The size of the moving average sample window may be, for example, three neighboring focus factors. For some focus profiles, smoothing may improve the accuracy of the local maximum search.

0091In Step S<b>320,</b> the computer system <b>106</b> identifies the image number corresponding to the local maximum identified in Step S<b>318</b> and extracts the image slice associated with that identified image number from the stack of tomosynthesis images for display in Step S<b>322.</b> The extracted image slice is also referred to herein as a high-focus image, because it has a greater focus factor (that is, it is in greater focus) than other nearby image slices, as determined in Step S<b>318.</b> In one example embodiment herein, the high-focus image is extracted and stored in the main memory <b>232,</b> although it may be stored instead in the secondary memory <b>234,</b> and can be retrieved therefrom for display in Step S<b>322.</b>

0092In the case where more than one local maximum is found in Step S<b>318,</b> more than one high-focus image corresponding to those local maxima are extracted. In an example embodiment herein, the high-focus images show one or more sub-object(s) <b>52</b> with clarity, by virtue of the processing performed in Steps S<b>314,</b> S<b>316,</b> and S<b>318,</b> even though the sub-object(s) <b>52</b> may not be visible in the conventional orthogonal projection image. This is because the presence of in-focus sub-object(s) <b>52</b> (e.g., anatomical features) in a high-focus image generally corresponds to a high mean variance in the region of interest of that high-focus image, and the high mean variance can be identified as a local maximum in Step S<b>320.</b> For example, <figref idref="f0012">FIGS. 6 and 7</figref> illustrate examples of image slices extracted by performing Step S<b>320.</b> First, local maxima <b>502</b> and <b>504</b> in <figref idref="f0011">FIG. 5</figref> are identified as corresponding to image numbers <b>7</b> and <b>36</b> in the tomosynthesis stack. In <figref idref="f0012">FIGS. 6 and 7</figref>, which correspond to image numbers <b>7</b> and <b>36,</b> respectively, clearly defined root structures <b>602</b> and <b>702</b> can be seen, where those structures appear more clearly than in the orthogonal projection image shown on <figref idref="f0010">FIG. 4</figref>.

0093In Step S<b>322,</b> the computer system <b>106</b> displays the high-focus image(s) extracted in Step S<b>320</b> on the display unit <b>108.</b> In one example embodiment herein, the display unit <b>108</b> displays a composite image comprising a portion of the high-focus image extracted in Step S<b>320,</b> overlaid on the orthogonal projection image extracted in Step S<b>308.</b> The composite image is formed by using the region of interest received in Step S<b>312</b> to select a corresponding portion of the high-focus image extracted in Step S<b>320.</b> The portion of the high-focus image is then overlaid on the region of interest indicated in Step S<b>312</b> of the orthogonal projection image extracted in Step S<b>308.</b> For example, in the example embodiment described above in Step S<b>304</b> where each image slice and the orthogonal projection image have the same dimensions, position, and orientation, the region of interest indicated in Step S<b>312</b> corresponds to the same pixel locations in both the high-focus images extracted in Step S<b>320</b> and the orthogonal projection image extracted in Step S<b>308.</b> In the case where more than one high-focus image is extracted in Step S<b>320,</b> more than one composite image can be displayed in Step S<b>322.</b> For example, <figref idref="f0013">FIGS. 8 and 9</figref> illustrate examples of the orthogonal projection image shown in <figref idref="f0010">FIG. 4</figref> with the region of interest portions from the high-focus images of <figref idref="f0012">FIGS. 6 and 7</figref>, respectively, overlaid thereon.

0094In another example embodiment herein, displaying the high-focus image(s) in Step S<b>322</b> can also include displaying a user interface on display unit <b>108,</b> with which the user may interact via input unit <b>114</b> and/or display unit <b>108.</b><figref idref="f0014">FIG. 10</figref> illustrates an example of such a user interface <b>1000.</b> The user interface <b>1000</b> can display an orthogonal projection image in an orthogonal projection image viewport <b>1002.</b> A region of interest <b>1006</b> can be visually depicted in the orthogonal projection image viewport <b>1002</b> and can be the same region of interest obtained from the user in Step S<b>312.</b>

0095The user interface <b>1000</b> can also display, in a tomosynthesis image viewport <b>1004,</b> the entire stack of tomosynthesis image slices, or one or more selected image slices of the stack. Stack controls <b>1016</b> can be provided to enable the user to manually select, via a scroll bar for example, which image slice of the tomosynthesis stack to display on the tomosynthesis image viewport <b>1004.</b> In another example embodiment herein, the stack controls <b>1016</b> may include selectable control items, such as, for example, play, pause, skip forward, and skip backward, (not shown), or the like, to enable the user to control automatic display of the tomosynthesis stack, as a cine loop for example, on the tomosynthesis image viewport <b>1004.</b> An image location indicator <b>1018</b> also can be provided to indicate to where in object <b>50</b> along the z-axis relative to the x-ray detector <b>102</b> the image slice appearing in the tomosynthesis image viewport <b>1004</b> corresponds. The location along the z-axis of the image slice appearing in the tomosynthesis image viewport <b>1004,</b> represented by the image location indicator <b>1018,</b> is known, because, as described above, the location of each image slice in the stack of tomosynthesis image slices reconstructed in Step S<b>304</b> was provided as an input to the tomosynthesis image reconstruction process of that step. In an alternative embodiment herein, the z-axis location represented by the image location indicator <b>1018</b> can be calculated (e.g., by the computer system <b>106</b> or the computer processor <b>222</b>) by multiplying the image number of the image appearing in the tomosynthesis image viewport <b>1004</b> by the z-axis slice thickness, which is a function of the known geometry of the system (e.g., the scan angle <b>112</b>).

0096The user interface <b>1000</b> can also include pictorial indicators to help the user navigate the stack of tomosynthesis image slices, such as, for example, a representation of the imaged object <b>1008</b> (which is a pictorial representation of object <b>50</b>), an x-ray detector icon <b>1010</b> (placed in relation to the representation of the imaged object <b>1008</b> to indicate the placement of x-ray detector <b>102</b> relative to object <b>50</b> in Step S<b>302</b>), and high-focus image indicator lines <b>1012</b> and <b>1014</b> within the imaged anatomy <b>1008</b> (which are pictorial representations that indicate the z-axis location within object <b>50</b> of high-focus images identified and extracted in Step <b>S320</b>).

0097The number of high-focus image indicator lines appearing on the user interface <b>1000</b> corresponds with the number of high-focus images extracted in Step S<b>320.</b> For example, in <figref idref="f0014">FIG. 10</figref>, two high-focus image indicator lines <b>1012</b> and <b>1014</b> are shown, representing two high-focus images extracted in Step S<b>320</b> and illustrated in the examples of <figref idref="f0012">FIGS. 6 and 7</figref>, respectively. In one example embodiment herein, selecting a high-focus image indicator line will display the corresponding high-focus image identified and extracted in Step S<b>320</b> in the tomosynthesis image viewport <b>1004.</b>

0098The process of <figref idref="f0009">FIG. 3</figref> ends at Step S<b>324.</b>

0099Clinician users are accustomed to reading conventional two-dimensional x-ray images due to their long-established use in medicine, but 3D datasets may be less familiar and more difficult to analyze. It can be appreciated that automated extraction of information from a 3D tomographic dataset, including the identification within the 3D tomographic dataset of high-focus images that contain features of interest (e.g., anatomical features), and presentation of the extracted information to a user may be useful. A clinician user may find it more intuitive to have a portion of the high-focus image, which shows the feature of interest, overlaid on and within the context of the two-dimensional orthogonal projection image, with which they are accustomed from long-established practice. Additionally, automated presentation of the extracted information can save the clinician user the time and effort associated with manually scrolling through and analyzing a large number of image slices in order to locate features of interest.

Reducing Image Reconstruction Artifacts

0100The above intraoral tomosynthesis system <b>100</b> will now be further described in conjunction with <figref idref="f0021">FIG. 17A</figref>, which shows a flow diagram of a process according to an example embodiment herein for using a mask based on a two-dimensional orthogonal projection image to guide a process for reducing image reconstruction artifacts in an intraoral tomosynthesis dataset.

0101Prior to starting the process, the x-ray detector <b>102</b> and x-ray source <b>104</b> are aligned manually by a user to a starting position, as described above, in one example embodiment herein.

0102The process of <figref idref="f0021">FIG. 17A</figref> starts at Step S<b>1700,</b> and in Step S<b>1702</b> the intraoral tomosynthesis system <b>100</b> acquires a plurality of projection images of object <b>50</b> over a scan angle <b>112</b> (which may be predetermined), including the orthogonal projection image, in the manner described above. For example, the x-ray source <b>104</b> is moved by the motorized stage <b>116</b> and control circuitry <b>118</b> to different positions within the scan angle <b>112,</b> and the computer system <b>106</b> controls the x-ray source <b>104</b> to emit x-rays <b>110</b> at each position. In one example embodiment herein, x-ray source <b>104</b> is scanned, by pivoting at a point along the z-axis, from -20° from the z-axis to +20° from the z-axis in evenly distributed increments of 0.8° to provide <b>51</b> scan angles, including the 0° position, although this example is not limiting. The x-rays <b>110</b> then pass through and are attenuated by the object <b>50</b> before being projected onto the x-ray detector <b>102.</b> The x-ray detector <b>102</b> converts the x-rays <b>110</b> into electrical signals (either directly or indirectly, as described above) and provides the electrical signals to the computer system <b>106.</b> The computer system <b>106</b> processes the electrical signals collected at each scan angle position to acquire the plurality of projection images, each image comprising an array of pixels. The image acquired with the x-ray source <b>104</b> at the 0° position is also referred to herein as an orthogonal projection image.

0103In one example embodiment herein, the color depth of each pixel value of the projection images may be 12-bit grayscale, and the dimensions of the projection images correspond to the standard dental size of the x-ray detector <b>102,</b> as described above. For example, a Size-2 detector may produce projection images that are approximately 1700×2400 pixels in size, a Size-1 detector may produce projection images that are approximately 1300×2000 pixels in size, and a Size-0 detector may produce projection images that are approximately 1200×1600 pixels in size.

0104In Step S<b>1704,</b> the computer system <b>106</b> processes the plurality of projection images acquired in Step S<b>302</b> using a reconstruction technique in order to reconstruct a series of two-dimensional tomosynthesis image slices and may also perform deblurring and other image enhancements, as will be described further herein. Each reconstructed image slice is a tomographic section of object <b>50</b> comprising an array of pixels (each pixel being located at a pixel coordinate), that is, each image slice represents a cross-section of object <b>50</b> that is parallel to the x-y plane in which the receiving surface of the x-ray detector <b>102</b> extends, has a slice thickness along the z-axis, and is positioned at a different, respective location along the z-axis than other image slices. The slice thickness is a function of the reconstruction technique and aspects of the geometry of the system <b>100,</b> including, primarily, the scan angle <b>112.</b> For example, each image slice may have a slice thickness of 0.5 mm by virtue of the geometry of the system <b>100</b> and the reconstruction technique. The desired location of each reconstructed image slice along the z-axis is provided as an input to the reconstruction performed in Step S<b>304</b> either as a pre-programmed parameter in computer system <b>106</b> or by user input via input unit <b>114</b> and/or display unit <b>108.</b> By example only, the computer system <b>106</b> can be instructed to reconstruct, from the plurality of projection images, a first image slice that is one millimeter (1 mm) away from the surface of x-ray detector <b>102</b> along the z-axis, a last image slice being at fifteen millimeters (15 mm) away from the surface of the x-ray detector <b>102,</b> and image slices between the first image slice and the last image slice at regular increments along the z-axis of two-hundred micrometers (200 µm), for a total of seventy-one image slices.

0105Reconstruction of the tomosynthesis image slices in Step S<b>1704</b> may be performed in accordance with any existing or later developed reconstruction technique. In one example embodiment herein, reconstruction of the tomosynthesis image slices in Step S<b>1704</b> utilizes a shift-and-add technique, described above. The shift-and-add technique utilizes information about the depth of sub-object(s) <b>52</b> along the z-axis that is reflected in the parallax captured by the plurality of projection images, as described above. According to this example embodiment, an image slice is reconstructed by first spatially shifting each projection image by an amount that is geometrically related to the distance between the image slice and the tomographic focal spot <b>122</b> along the z-axis. The shifted projection images are then averaged together to result in the image slice, where all sub-objects <b>52</b> in the plane of the image slice are in focus and sub-objects <b>52</b> outside of that plane are out of focus and blurry. This shift-and-add process is repeated for each image slice to be reconstructed. In the case of the image slice corresponding to the x-y plane that includes the tomographic focal spot <b>122,</b> the projection images are averaged together without first shifting because sub-objects <b>52</b> are already in focus for that plane.

0106The foregoing describes a basic shift-and-add reconstruction technique. In one example embodiment herein, a deblurring technique that substantially reduces or removes blurry, out-of-plane sub-objects from an image slice can be performed in conjunction with the reconstruction technique (whether shift-and-add or another technique). Examples of deblurring techniques that can be employed include, for example, spatial frequency filtering, ectomography, filtered backprojection, selective plane removal, iterative restoration, and matrix inversion tomosynthesis, each of which may be used in Step S<b>1704</b> to deblur images reconstructed by the shift-and-add reconstruction technique (or another reconstruction technique, if employed).

0107In another example embodiment herein, Step S<b>1704</b> also can include the computer system <b>106</b> performing further automated image enhancements such as, for example, image blurring, image sharpening, brightness optimization, and/or contrast optimization, on each reconstructed (and deblurred, where deblurring is performed) image slice in a known manner.

0108Additionally, in another example embodiment herein, the dimensions, position, and orientation of each image slice reconstructed in Step S<b>1704</b> are the same as the corresponding characteristics of the orthogonal projection image. Thus, when tomosynthesis image slices (or portions thereof) and the orthogonal projection image are overlaid over one another, corresponding anatomical features appearing in the images will be overlapped and aligned without scaling, rotation, or other transformation of the images.

0109After Step S<b>1704,</b> control passes to Step S<b>1716,</b> which will be described below. Before describing that step, Steps S<b>1706,</b> S<b>1708,</b> and S<b>1714</b> will first be described. Like Step S<b>1704,</b> Step S<b>1706</b> is performed after Step S<b>1702</b> is performed. (However, in other embodiments herein, Step S<b>1704</b> is performed serially between Steps S<b>1702</b> and S<b>1706,</b> and then the method continues from Step S<b>1706</b> as shown in <figref idref="f0021">FIG. 17A</figref>. That is, control passes from Step S<b>1702</b> to Step S<b>1704,</b> and then from Step S<b>1704</b> to Step S<b>1706,</b> in which case, control does not pass from Step S<b>1704</b> to Step S<b>1716</b> as shown in <figref idref="f0021">FIG. 17A</figref>. In yet other example embodiments herein, Step S<b>1704</b> is performed serially between Steps S<b>1714</b> and Step S<b>1716,</b> instead of the order shown in <figref idref="f0021">FIG. 17A</figref>. That is, control passes from Step S<b>1702</b> to Step S<b>1706,</b> at which point the method is performed until Step S<b>1714</b> according to <figref idref="f0021">FIG. 17A</figref>, then control passes from Step S<b>1714</b> to Step S<b>1704,</b> and then control passes from Step S<b>1704</b> to Step S<b>1716,</b> but control does not pass from Step S<b>1702</b> to Step S<b>1704</b> as shown in <figref idref="f0021">FIG. 17A</figref>.)

0110Referring again to the embodiment represented in <figref idref="f0021">FIG. 17A</figref>, in Step S<b>1706,</b> the orthogonal projection image is extracted from the plurality of projection images acquired in Step S<b>1702.</b> Because, as described above, the orthogonal projection image is defined as the projection image captured while the x-ray source <b>104</b> is in the 0° scan angle position, no reconstruction is necessary to extract that image. In one example embodiment herein, the orthogonal projection image is extracted and stored in the main memory <b>232,</b> although it may be stored instead in the secondary memory <b>234.</b> For example, <figref idref="f0024">FIG. 18</figref> illustrates an example orthogonal projection image.

0111In Step S<b>1708,</b> the orthogonal projection image extracted in Step S<b>1706</b> is received as input and processed to generate as output a verified list of pixel coordinates of the orthogonal projection image that do not contain a predetermined type of information, e.g., anatomic or diagnostic information (such regions are hereinafter referred to as "non-anatomic regions" for convenience). Non-anatomic regions correspond to, in the imaged volume, air gaps, air pockets, and/or a mechanism for holding the x-ray detector <b>102,</b> for example.

0112Step S<b>1708</b> can be performed in different ways according to various example embodiments herein, including, but not limited to, either of the subprocesses illustrated in <figref idref="f0022">FIGS. 17B</figref> and <figref idref="f0023">17C</figref>. The subprocesses shown in <figref idref="f0022">FIGS. 17B</figref> and <figref idref="f0023">17C</figref> will now be described in turn.

0113In one example embodiment herein, Step S<b>1708</b> (<figref idref="f0021">FIG. 17A</figref>) may be performed according to the subprocess illustrated in <figref idref="f0022">FIG. 17B</figref>. The subprocess of <figref idref="f0022">FIG. 17B</figref> starts in Step S<b>1730,</b> object edges within the orthogonal projection image extracted in Step S<b>1706</b> are detected by deriving at least one edge map from the orthogonal projection image and comparing the edge map to a threshold, in a manner described below. The edge map can be a variance image, a gradient image, or a combination thereof, and, in one example, the variance image and the gradient image can be derived by applying, respectively, a variance kernel operator and a gradient kernel operator to the orthogonal projection image extracted in Step S<b>1706</b> in a manner described in greater detail below. The sensitivity of the variance and gradient kernel operators to edges, and the width of the detected edges on the variance and gradient images, can be adjusted by tuning the parameters of the operators. Additionally, after deriving the edge map but prior to comparing the edge map to a threshold, Step S<b>1730</b> may also include a sub step of closing incomplete edges within the edge map, as will be described further below. Object edges detected in Step S<b>1730</b> are characterized by a high spatial frequency (i.e., a sharp transition) and are typically associated with boundaries of objects of interest in an image.

0114A variance image (mentioned above) can be derived in Step S<b>1730</b> by iterating a variance kernel operator, for example, a 5×5 pixel matrix, through each pixel coordinate of the orthogonal projection image. At each iterative pixel coordinate, the statistical variance of pixel values of the orthogonal projection image within the variance kernel operator is calculated, and the result is assigned to a corresponding pixel coordinate in the variance image. Because variance represents the degree to which the value of a pixel varies from those of its neighbors, high spatial frequency regions, such as an object edge, have a high variance, while smooth, low spatial frequency regions have a low variance. For example, <figref idref="f0025">FIG. 19</figref> illustrates a variance image derived from the orthogonal projection image illustrated in <figref idref="f0024">FIG. 18</figref>.

0115A gradient image (mentioned above) can be derived in Step S<b>1730</b> by iterating a gradient kernel operator through each pixel coordinate of the orthogonal projection image. At each iterative pixel coordinate, the gradient kernel operator is convolved with pixel values of the orthogonal projection image at that pixel coordinate, and the result is assigned to a corresponding pixel coordinate in the gradient image. Examples of well-known gradient kernel operators include Roberts, Prewitt, and Sobel operators.

0116Because the gradient represents the magnitude of directional change in pixel value at a given pixel relative to its neighbors, a region with a greater directional change in pixel value, such as an object edge, will have a greater corresponding gradient value than a region with no change or non-directional change. For example, <figref idref="f0026">FIG. 20</figref> illustrates a gradient image derived from the orthogonal projection image illustrated in <figref idref="f0024">FIG. 18</figref>.

0117In one example embodiment herein, the variance image and the gradient image may individually serve as edge maps, and may be derived simultaneously or sequentially. In yet other example embodiments herein, only one of the variance image and the gradient image serves as an edge map. In a further example embodiment herein, the edge map is derived by combining the variance image and the gradient image by way of a logic operator (e.g., AND, OR, etc.).

0118After the edge map is derived in Step S<b>1730,</b> Step S<b>1730</b> may, in one example embodiment herein, further include a substep of closing incomplete edges within the edge map (also known as edge linking). The edge map may have incomplete edges due to a number of factors, including, for example, noise or discontinuities in the pixel values of the orthogonal projection image, the parameters of the variance and gradient kernels, and/or the threshold applied to the edge map. Incomplete edges can be closed according to mathematical techniques and/or localized curve fitting methods.

0119Finally, edges are extracted in Step S<b>1730</b> by identifying pixels in the edge map that are greater than a threshold. The value of the threshold may depend on whether the edge map corresponds to the variance image, the gradient image, or a logical combination thereof. Furthermore, adjusting the threshold allows for control over the sensitivity to edges and the width of the detected edges, and thus it also is within the scope of the method of <figref idref="f0021">FIGS. 17A</figref> and <figref idref="f0022">17B</figref> to employ an adjustable threshold in the edge extraction, wherein the threshold is adjusted based on predetermined criteria over iterations of edge extractions.

0120After Step S<b>1730</b> is performed, control passes to Step S<b>1734,</b> which will be described below. Before describing that step, Step S<b>1732</b> will first be described. Like Step S<b>1730,</b> Step S<b>1732</b> is performed after Step S<b>1706</b> (of <figref idref="f0021">FIG. 17A</figref>) is performed. (However, in other embodiments herein, Steps S<b>1730</b> and S<b>1732</b> can be performed serially. For example, instead of the order shown in <figref idref="f0022">FIG. 17B</figref>, in another example embodiment herein, Step S<b>1730</b> is performed before Step S<b>1732</b> (i.e., in between steps S<b>1706</b> and S<b>1732</b>), whereas in another example embodiment herein, Step S<b>1732</b> is performed before Step S<b>1730</b> (i.e., in between steps S<b>1706</b> and S<b>1730</b>)).

0121In Step S<b>1732,</b> image statistics of the orthogonal projection image (extracted in Step S<b>1706</b>) as a whole, including, for example, a mean pixel value, a median pixel value, a pixel value histogram, and a pixel value cumulative histogram, are calculated and used to estimate the locations of non-anatomic regions of the orthogonal projection image. As mentioned above, non-anatomic regions correspond to, in the imaged volume, air gaps, air pockets, and/or a mechanism for holding the x-ray detector <b>102,</b> for example. Because no anatomy is present in these regions to absorb emitted x-rays <b>110,</b> the detector <b>102</b> receives more x-rays <b>110</b> and outputs higher electrical signal in these regions, which appear, for example, as black or very dark areas on the projection images acquired in Step S<b>1702.</b> Thus, on the pixel value histogram, these non-anatomic regions have high pixel values relative to the rest of the orthogonal projection image, and may further correspond to one or more distinctive peaks at the high pixel values (such distinctive peaks in the pixel value histogram will correspond to large gradient increases in the pixel value cumulative histogram). The image statistics (e.g., the mean pixel value, the median pixel value, and the pixel value cumulative histogram) can be combined to determine a threshold point that separates pixel values corresponding to anatomy from pixel values corresponding to non-anatomy. For example, the threshold point may be specified manually by a user (via input unit <b>114</b> and/or display unit <b>108</b>) or automatically by computer system <b>106</b> by selecting a pixel value that is above the median pixel value but below the distinctive peak(s). Based on the threshold point, a range of non-anatomic pixel values is deemed to correspond to regions containing no anatomic information. Finally, Step S<b>1732</b> generates as an output a list of unverified non-anatomic pixel coordinates, which represents the estimated locations of non-anatomic regions of the orthogonal projection image, from the pixel coordinates of the orthogonal projection image (from Step S<b>1706</b>) that have pixel values within the range of non-anatomic pixel values.

0122For example, <figref idref="f0027">FIG. 21</figref> illustrates image statistics of the orthogonal projection image of <figref idref="f0024">FIG. 18</figref>, including a pixel value histogram <b>2100,</b> a pixel value cumulative histogram <b>2102,</b> a median of the cumulative histogram <b>2104,</b> a distinctive histogram peak at high pixel values <b>2106,</b> and a large gradient increase in the cumulative histogram at high pixel values <b>2108.</b> A threshold point <b>2110</b> is determined for the example illustrated in <figref idref="f0027">FIG. 21</figref> to be a pixel value of approximately 3800 based on the image statistics, and a range of non-anatomic pixel values of approximately 3800 and above is deemed to correspond to non-anatomic regions. Finally, Step S<b>1732</b> outputs an unverified list (not shown) of non-anatomic pixel coordinates from the orthogonal projection image of <figref idref="f0024">FIG. 18</figref> that have pixel values equal to or greater than 3800.

0123In certain circumstances, the list of unverified non-anatomic pixel coordinates generated in Step S<b>1732</b> may imperfectly separate anatomy from non-anatomy because the threshold point determined in Step S<b>1732</b> is based solely on the pixel value statistical information over the entire orthogonal projection image. That is, it may occur that some regions of the orthogonal projection image having pixel values within the range of non-anatomic pixel values (selected in Step S<b>1732</b> based on the threshold point) do in-fact contain anatomic information.

0124Thus, in Step S<b>1734,</b> the computer system <b>106</b> checks which pixel coordinates of the list of unverified non-anatomic pixel coordinates generated in Step S<b>1732</b> are inside a spatial boundary formed by the edges detected in Step S<b>1730</b> in order to generate a list of verified non-anatomic pixel coordinates. In particular, pixel coordinates within the boundaries are deemed to be anatomic regions, while pixel coordinates outside the boundaries are deemed to be non-anatomic regions. Accordingly, pixel coordinates determined to be inside the boundaries (i.e., anatomic regions) are removed from the list, and pixel coordinates determined to be outside of the boundaries (i.e., non-anatomic regions) remain on the list, thereby providing a list of verified non-anatomic pixel coordinates that is provided to Step S<b>1714</b> (<figref idref="f0021">FIG. 17A</figref>).

0125In another example embodiment herein, Step S<b>1708</b> may be performed according to the subprocess illustrated in <figref idref="f0023">FIG. 17C</figref>, which evaluates each pixel coordinate of the orthogonal projection image extracted in Step S<b>1706</b> according to a conditional statement that states that if a pixel coordinate has a high pixel value and a low pixel variance, then that pixel coordinate contains substantially no anatomic information.

0126In particular, the subprocess of <figref idref="f0023">FIG. 17C</figref> iterates through each pixel coordinate of the orthogonal projection image extracted in Step S<b>1706,</b> starting with a first pixel coordinate in Step S<b>1740.</b> Then, in Step S<b>1742,</b> the pixel variance at the starting pixel coordinate is calculated, for example, by computing the statistical variance of pixel values of the orthogonal projection image within a variance kernel operator (e.g., a 5×5 pixel matrix centered on the starting pixel coordinate).

0127In decision block S<b>1744,</b> the computer system <b>106</b> evaluates whether the pixel variance calculated in Step S<b>1742</b> is low (by comparing the pixel variance to a pixel variance threshold, which may be predetermined) and whether the pixel value at the pixel coordinate is high (by comparing the pixel coordinate to a pixel value threshold, which may be predetermined). If both conditions are satisfied in decision block S<b>1744,</b> then, in Step S<b>1746,</b> the pixel coordinate is added to a verified list of pixels deemed to contain substantially no anatomic information before control passes to decision block S<b>1748.</b> If either condition is not satisfied in decision block S<b>1744,</b> then control passes directly to decision block S<b>1748.</b>

0128A pixel coordinate satisfying both conditions is deemed to contain substantially no anatomic information because of the following presumptions regarding pixel variances and pixel values of the orthogonal projection image. First, areas of the orthogonal projection image containing anatomic information generally exhibit at least a moderately high pixel variance, while areas not containing anatomic information (e.g., air gaps, air pockets, or an x-ray detector holder) generally exhibit a low pixel variance. Second, as explained above with respect to Step S<b>1732,</b> pixel coordinates that contain no anatomic generally have high pixel values and appear black or very dark, because no anatomy is present to absorb the emitted x-rays <b>110.</b> On the other hand, pixel coordinates containing anatomic information generally exhibit lower pixel values. Thus, if a pixel coordinate has both low pixel variance and high pixel value, it is deemed likely to not contain anatomic information.

0129In some example embodiments herein, the pixel value threshold used in decision block S<b>1744</b> can be derived from image statistics of the orthogonal projection image extracted in Step S<b>1706</b> as a whole, as described above with respect to Step S<b>1732</b> of the subprocess illustrated in <figref idref="f0022">FIG. 17B</figref>.

0130At decision block S<b>1748</b> (after decision block S<b>1744</b> if "No" at decision block S<b>1744</b> or after Step S<b>1746</b> if "Yes" at decision block S<b>1744</b>), the computer system <b>106</b> checks if the last pixel coordinate in the orthogonal projection image has been evaluated by the conditional statement at decision block S<b>1744.</b> If the last pixel coordinate has not been evaluated yet ("No" at decision block S<b>1748</b>), then the subprocess continues to Step S<b>1750</b>, where the computer system 106 iterates to the next pixel coordinate and subsequently evaluates that next pixel coordinate in the above described manner (i.e., performs Steps S<b>1742</b>, S<b>1744</b>, S<b>1746</b> if applicable, and S<b>1748</b>). If the last pixel coordinate has been evaluated ("Yes" at decision block S<b>1748</b>), then the subprocess of <figref idref="f0023">FIG. 17C</figref> ends by passing control and the verified list of pixels compiled in Step S<b>1746</b> to Step S<b>1714</b> (<figref idref="f0021">FIG. 17A</figref>).

0131It will be apparent to a person skilled in the relevant art(s) that while the subprocess of <figref idref="f0023">FIG. 17C</figref> is explained above on a pixel-by-pixel basis, the subprocess of <figref idref="f0023">FIG. 17C</figref> also can be performed by different computational techniques, including, for example, evaluating the conditional statement for the whole orthogonal projection image extracted in Step S<b>1706</b> in a single matrix operation based on the orthogonal projection image and a variance image generated from the orthogonal projection image (e.g., a variance image can be derived as described above at Step S<b>1730</b>).

0132Referring again to <figref idref="f0021">FIG. 17A</figref>, in Step S<b>1714,</b> a binary mask image, having the same dimensions as the orthogonal projection image, is formed based on the list of verified non-anatomic pixel coordinates provided in Step S<b>1708.</b> Each verified pixel coordinate on the binary mask image is assigned a pixel value of one. All other pixel coordinates on the binary mask image are assigned a pixel value of zero. For example, <figref idref="f0028">FIG. 22</figref> illustrates an example binary mask image obtained in Step S<b>1714</b> (and corresponding to the orthogonal projection image of <figref idref="f0010">FIG. 4</figref>).

0133In Step S<b>1716,</b> each tomosynthesis image slice reconstructed in Step S<b>1704</b> is masked by the binary mask image formed in Step S<b>1714</b> to isolate the regions with no anatomic information. In an example embodiment herein, Step S<b>1716</b> is performed by multiplying the binary mask image with each tomosynthesis image slice reconstructed in Step S<b>1704</b> to provide a corresponding masked tomosynthesis image slice. For example, <figref idref="f0029">FIG. 23</figref> illustrates a tomosynthesis image slice masked by the binary mask image of <figref idref="f0028">FIG. 22</figref>, resulting in an isolated nonanatomic region <b>2300.</b>

0134In Step S<b>1718,</b> the pixel values in the isolated nonanatomic regions of each masked tomosynthesis image slice obtained in Step S<b>1716</b> are compressed to provide corresponding reduced-artifact tomosynthesis image slices. In an example embodiment herein, such compression is achieved by mapping the pixel values in the isolated regions through an intensity transformation function into different output pixel values. The intensity transformation function may be, in one example, configured to assign very low contrast to pixel values in isolated regions like, for example, the intensity transformation function illustrated by graph <b>2400</b> of <figref idref="f0030">FIG. 24</figref>.

0135In addition to compressing pixel values in the isolated regions, Step S<b>1718</b> may also include a further substep of applying an anatomic-region intensity transformation function to increase the contrast in regions containing anatomic information, the anatomic-region intensity transformation function being different from the intensity transformation function applied to isolated regions. Examples of anatomic-region intensity transformation functions include a linear intensity transformation function illustrated by graph <b>2402</b> of <figref idref="f0030">FIG. 24</figref>, a sinusoidal intensity transformation function illustrated by graph <b>2500</b> of <figref idref="f0031">FIG. 25</figref>, and a logarithmic intensity transform function (not shown). In the case that the foregoing substep is performed, Step S<b>1718</b> outputs reduced-artifact tomosynthesis image slices where the pixel values in isolated regions have been compressed and pixel values in anatomic-regions have been contrast-enhanced.

0136In Step S<b>1720,</b> the computer system <b>106</b> displays one or more of the reduced-artifact image slices obtained in Step S<b>1718</b> (i.e. the slices corresponding to respective tomosynthesis image slices reconstructed in Step S<b>1704</b>) on the display unit <b>108.</b> In one example embodiment herein, the displaying can be performed so as to show all reduced-artifact image slices, or one or more selected reduced-artifact image slices, using display unit <b>108,</b> and interactive controls are provided (e.g., via display unit <b>108</b> and/or input device <b>114</b>) to enable a user to select between those two options, and to select one or more reduced-artifact image slices for display, and also to select one or more particular regions of interest in the reduced-artifact image slice(s) for display (whether in zoom or non-zoom, or reduced fashion). For example, <figref idref="f0032">FIG. 26</figref> illustrates an example reduced-artifact image slice that may be displayed on the display unit <b>108.</b>

0137In other example embodiments herein, reduced-artifact image slices obtained in Step S<b>318</b> may also be utilized in further processing methods (not shown) with or without being displayed in Step S<b>320.</b> In one example embodiment herein, the reduced-artifact image slices can be utilized to generate clinical information from a tomosynthesis dataset according to the method described further herein below with reference to <figref idref="f0008">FIG. 2B</figref>. In another example embodiment herein, the reduced-artifact image slices can be utilized to interactively extract features for intuitive presentation to a clinician user, as described above. In another embodiment example, the reduced-artifact image slices can be utilized to render a 3D image of object <b>50</b> and/or sub-object(s) <b>52.</b> In a further embodiment example herein, the reduced-artifact image slices can be utilized for measuring distances between points or structures of object <b>50</b> and/or sub-object(s) <b>52.</b>

0138The process of <figref idref="f0021">FIG. 17A</figref> ends at Step S<b>1722.</b>

Three-Dimensional Image Rendering from Image Slices

0139<figref idref="f0033">FIGS. 27A</figref> and <figref idref="f0034">27B</figref> are flow diagrams of processes for rendering a three-dimensional (3D) image from tomosynthesis slices according to example embodiments herein, or a sequence that can be used to enable such rendering.

0140Briefly, in <figref idref="f0033">FIGS. 27A</figref> and <figref idref="f0034">27B</figref>, a series of two-dimensional x-ray projection images over a limited scan angle are obtained. Tomosynthesis slices are reconstructed from the x-ray projection images. A region-of-interest is defined in one or more of the tomosynthesis slices. An outline trace of one or more objects in the region-of-interest is created for each of the slices. Following this, a volume to be rendered may be generated by steps illustrated in either <figref idref="f0033">FIG. 27A</figref> or <figref idref="f0034">27B</figref>, as discussed in more detail below.

0141As shown in <figref idref="f0033">FIGS. 27A</figref> and <figref idref="f0034">27B</figref>, in step <b>2701,</b> x-ray projection images are acquired over a limited scan angle. As discussed above, at each imaging position within the scan angle, x-ray source <b>104</b> emits x-rays <b>110</b> which pass through object <b>50</b> and are detected by the x-ray detector <b>102.</b> For each imaging position, a projection image is obtained by the computer system <b>106</b> based on the intensity of the x-rays received by x-ray detector <b>102.</b> Thus, the system <b>100</b> collects a plurality of projection images (also referred to herein as "projections") by positioning the x-ray source <b>104</b> at different angles, including, for example, the 0° position, and emitting x-rays <b>110</b> at each of those different angles through object <b>50</b> towards x-ray detector <b>102.</b>

0142In step <b>2702</b> of <figref idref="f0033">FIGS. 27A</figref> and <figref idref="f0034">27B</figref>, a series of tomosynthesis slices are reconstructed by computer system <b>106</b> from the plurality of projections images to form a tomosynthesis image stack. Each tomosynthesis slice is parallel to a plane in which the front surface of the x-ray detector <b>102</b> extends and at a particular depth in the z-axis direction corresponding to that tomosynthesis slice. Image reconstruction may be accomplished by several different reconstruction techniques including, for example, a shift-and-add method, filtered backprojection, matrix inversion tomosynthesis, generalized filtered backprojection, SIRT (simultaneous iterative reconstruction technique), or algebraic technique, among others.

0143In an exemplary embodiment, a shift-and-add method may be used. As discussed, the shift-and-add method takes into account the fact that objects at different heights (relative to the detector) will undergo varying degrees of parallax when exposed to x-rays at varying angles. To develop an image of objects at a specific height (i.e., a reconstructed tomosynthesis slice corresponding to a specific height), each projection image is shifted and added together with the other projection images at that height from different angles such that all objects in a plane at that height are in focus and objects outside of that plane are out of focus.

0144In step <b>2703</b> of <figref idref="f0033">FIGS. 27A</figref> and <figref idref="f0034">27B</figref>, the reconstructed tomosynthesis slices obtained in step <b>2702</b> are preprocessed. The tomosynthesis slices may be preprocessed by undergoing one or more of edge enhancement, noise reduction filtering, and edge detection filtering. The preprocessing steps will be described more fully below with respect to <figref idref="f0036">FIG. 28</figref>.

0145In step <b>2704</b> of <figref idref="f0033">FIGS. 27A</figref> and <figref idref="f0034">27B</figref>, a region-of-interest in at least one of the slices that underwent preprocessing in step <b>2703</b> is selected for 3D rendering. In one example, an operator or user (such as a dentist) can operate an input unit (e.g., input unit <b>114</b>) to define a region-of-interest in one or more of the tomosynthesis slices in the tomosynthesis image stack, while a display unit (e.g., display unit <b>108</b>) displays the slice and the current region-of-interest. For example, an operator can drag a mouse to select a rectangle corresponding to the region-of-interest on a displayed 2-D tomosynthesis slice, as shown in <figref idref="f0037">FIG. 29</figref> and described more fully below. Of course, the invention is not limited to this procedure for selecting a region-of-interest. In another example, the region could be traced on a display screen, selected by entering coordinates, or could be automatically selected based on predefined criteria.

0146As shown in <figref idref="f0033">FIG. 27A</figref>, after the region-of-interest is selected, the volume gradient for the region-of-interest, in directions parallel to the plane of the sensor, is measured in step <b>2719.</b> From the measured volume gradient in step <b>2719,</b> a three-dimensional outline trace of the object may be determined (step <b>2720</b>) by a variety of techniques, including a gradient segmentation technique or a minimal path (geodesic) technique, as described below and illustrated in <figref idref="f0035">FIGS. 27C and 27D</figref>.

0147<figref idref="f0035">FIG. 27C</figref> illustrates exemplary steps in determining the outline trace using a gradient segmentation technique. In a gradient segmentation technique, the image volume is segmented (i.e., divided into sections) based on the magnitude of the gradient, and the output of the gradient segmentation is a binary image. The gradient is segmented to identify regions with a gradient magnitude (i.e., the absolute value of the gradient) above a certain threshold. Regions which are above the threshold are assigned one binary value (e.g., 1) and regions below the threshold are assigned another binary value (e.g., 0), resulting in a binary image. In step <b>2721,</b> the measured volume gradient is processed as described above to produce a binary image which is then analyzed to determine connected edges, i.e., places where regions of different binary values are adjacent to one another. By determining the connected edges, separate objects may be identified and segmented from one another.

0148Once the objects are segmented, one or more object metrics such as the centroid, length, gradient direction, orientation, and root edge position relative to the region-of-interest may be calculated for each object. Based on one or more of these metrics, in step <b>2722,</b> a fuzzy logic algorithm tailored to a specific root geometry for a tooth may be used to determine which edge of the connected edges of a segmented object is most likely to be a given part of the tooth. The specific root geometry may be selected from a library of root geometries stored in a computer. Fuzzy logic is a many-valued logic used to evaluate approximate reasoning. In this embodiment, fuzzy variables have values between zero (0) and one (1), inclusive. A value of zero (0) corresponds to false. A value of one (1) corresponds to true. Values between zero (0) and one (1) represent a probability of being true. For each relevant object metric, an acceptable range of values is identified and a measure of trueness is determined. For object metrics that are known to occupy a fixed range, for example, root edge position relative to the region-of-interest, the related fuzzy variable is zero outside of the range and a nonzero isosceles triangle within the range. More specifically, the probability increases linearly from one end of the range until a maximum probability is reached in the middle of the range, after which the probability linearly decreases until the end of the range is reached; thus forming an isosceles triangle with a maximum probability in the middle of the variable range. For object metrics that have a minimum value, the related fuzzy variable is zero below the minimum value and increases linearly until the object metric hits a defined saturation value. Each fuzzy variable corresponds to the probability that an object with a given metric value corresponds to a given portion of the tooth. The object metrics that are used are the maximum height, minimum height, the height, and the width of the object. To identify a given feature, the fuzzy variables are multiplied and the highest measure is selected. Alternately, a neural network may be used to identify the edges most likely to belong to the object.

0149In an alternative embodiment of embodiment 1, the outline trace is calculated using geodesic approach, that is by determining the quickest travel time between two points on a curved surface. First, a potential volume is estimated from the gradient volume calculated in step <b>2719.</b> In an exemplary embodiment, the potential volume may be estimated by selecting either positive or negative gradient values in the direction of source travel (step <b>2721</b>A). Each value in the gradient volume may be compared to a threshold value to identify regions that have relatively high gradient values, either positive or negative. Either positive or negative gradient values are selected, and the non-selected opposite gradient values are set to zero in step <b>2722</b>A<b>.</b> To avoid unreal values when the thresholded gradient volume is inverted, an offset is added to the zero values in step S<b>2723</b>A. The amount of offset is chosen to reduce the ratio of the highest and lowest values in the estimated travel time to below a predetermined value (e.g., 10:1 ratio between highest and lowest values). The thresholded gradient volume is then inverted and squared (S<b>2724</b>A) to produce a potential volume.

0150Once the potential volume has been calculated, paths of minimal travel are determined to identify edges of objects (S<b>2725</b>A). One exemplary technique for calculating the paths of minimal travel is a fast march method, however, other techniques may also be used. In a fast march method, a series of calculations are performed to determine the quickest travel time from a starting point to one of the adjacent points (point A). Then, another series of calculations are performed to determine the quickest travel time from point A to one of the adjacent points (point B). This process repeats until the travel path reaches the desired point. Here, the quickest travel path coincides with an edge of the object and thus may be used to identify the edges (as opposed to the techniques discussed above for steps <b>2721</b> and <b>2722</b>). This technique has the advantage of automatically identifying three-dimensional branches and being insensitive to small breaks in an edge. In a case where any spurious edges (i.e., edges that are not part of the tooth) are identified, those edges may be eliminated by reference to a nearby volume intensity, edge shape, or a known tooth topology. For example, a nearby volume intensity may be used to determine a difference in intensities between volumes on either side of the identified edge. If the edge corresponds to a tooth, then the difference in intensities should be above a certain threshold. If the difference is below that threshold, then the edge likely does not correspond to a tooth's edge and may be considered a spurious edge. In another example, if an identified edge bends abruptly then it is unlikely to be part of a real tooth. Furthermore, if the topology of the tooth is known then the identified edge may be compared to the known topology to determine whether the edge corresponds to the tooth.

0151Once the outline trace is determined using either of the techniques discussed above for embodiment 1, the outline trace is matched to corresponding points on a reference object model in step <b>2726</b>. The reference model object is a surface model of a representative object such as, for example, a molar with an appropriate number of roots. The reference model object may be obtained either from available digital databases or may be generated either by means of a commercial three-dimensional optical scanner or by means of an x-ray tomography system. Although the outline trace determines the spatial position of the edges of the object, the depth of the object may not be determined by the outline trace, as it is not visible in the limited angle reconstruction. The position in a depth direction of the front and back faces of the object are estimated based on the ratio of the width of the reference object model to the thickness of the reference object model (aspect ratio). Positions of points on the front and back of the object surface are determined using the reference object model aspect ratio and the width of the outline trace, so as to generate an estimated object with a similar aspect ratio to that of the reference object. In step <b>2727</b>, a morphing transformation based on the relationship between the model object and the estimated object may be applied to the model surface to generate the estimated surface. A morphing transformation is a method of mapping one volume onto another volume in a continuous fashion. In an exemplary embodiment, a locally bounded three-dimensional Hardy transform may be used as a morphing transformation. The three-dimensional Hardy transform may be estimated based on the corresponding points on the reference model object and the estimated object.

0152Next, in step <b>2728</b>, the estimated surface created above is converted to a volume model and is reprojected into images matching the size and dimension of the original projections while assuming nominal attenuation values for dentin/enamel/pulp/air. In one embodiment, the estimated surface may be converted to a volume model first by assigning points adjacent to the estimated surface to the volume model and then filling the resulting hollow volume model. This may be accomplished using projection matrices calculated to match the projection geometry and using a calculation technique similar or identical to techniques used to estimate the backprojection matrices used for reconstruction in order to ensure the generation of compatible artifacts in the reprojected volume. Once the projection matrices are calculated, each projection matrix is then multiplied by the volume model in a matrix multiplication operation to obtain a projection of the volume model. The number of two-dimensional projections corresponds to the number of projection matrices. The two-dimensional projection images obtained by reprojecting the volume model obtained above (i.e., by simulating the passage of x-rays through the volume model) are subtracted from the measured projections and the resulting images are inspected to determine whether or not the estimated volume model is consistent with the data. Regions of the two-dimensional projections that contain accurately calculated objects are expected to be substantially uniform and non-negative. As such, regions within the two-dimensional projections that are substantially non-uniform or negative indicate inaccuracy in the volume model. Such regions may be determined by a thresholding operation. More specifically, the difference between the calculated projection and the recorded projection may be calculated, and for a given pixel if the difference exceeds a threshold value, that pixel is determined to be a point of inaccuracy. The model surface is then adjusted by adjusting the position of the extrapolated points (the points on the front and back of the object surface) in order to improve the consistency of the model with the measured data. For instance, if the remaining attenuation in the subtracted projections is less than zero, the volume model thickness decreases. This is accomplished by local adjustment of the aspect ratio of the reference model used to calculate the model front and back surfaces.

0153The technique described above, is also applied to the pulp chamber and root canal. In step <b>2714,</b> the resulting images are displayed or otherwise presented.

0154Alternatively, in embodiment 2 (<figref idref="f0034">FIG. 27B</figref>), after a region-of-interest in at least one of the slices that underwent preprocessing in step <b>2703</b> is selected for 3D rendering, corresponding region-of-interests ("ROI") are received by computer system <b>106</b> for each of the other slices in the tomosynthesis stack. In one example embodiment herein, the stack of reconstructed slices can be searched to identify those that are in focus for a region-of-interest selected in a single slice. As noted above, the tomosynthesis stack is composed of a series of tomosynthesis slices that are two-dimensional images in respective planes that are parallel to the surface of the x-ray detector. In an exemplary embodiment, a user may scroll through the tomosynthesis stack. For example, a user can operate an input unit (e.g., input unit <b>114</b>) to "scroll" or "page" through each, or selected ones, of the tomosynthesis slices in the tomosynthesis stack, while a display unit (e.g., display unit <b>108</b>) displays those slices. To enable viewing of the slices, the displayed tomosynthesis stack can include a manual scroll bar and/or automation controls such as play, pause, stop, rewind or the like. As such, if the user is interested in focusing on a particular region-of-interest, or in determining the depth placement of particular features in an image, the user may interact with such functions to scroll through the tomosynthesis stack and visually evaluate each image or selected images (e.g., in order to designate a corresponding region-of-interest in each of the images).

0155In step <b>2706,</b> a loop is started to loop through each slice of the stack of N slices, e.g., slice "1" to slice "N".

0156In step <b>2707,</b> a "shrink-wrapping" algorithm is applied around the structure in the current slice. Generally, a shrink-wrapping algorithm approximates a surface by starting with a triangular mesh and deforming the mesh to transform it to the required surface. A "shrink-wrapping" algorithm is a variant of an active contour/snake algorithm. An active contour/snake algorithm operates by establishing a set of grid points that form a contour enveloping the object and progressively evolving the positions of the grid points subject to rules based on the image content. For example, for a "shrink-wrapping" style operation, the initial grid positions are placed outside of the object and these positions are evolved by rules assuming a repulsive "force" associated with the object and an attractive force between adjacent grid points. Thus, according to one example embodiment herein, the outline trace of the region-of-interest in each slice is created by applying a shrink-wrapping algorithm.

0157At the same time (or in another embodiment, not at the same time), in step <b>2708,</b> an orthogonal input image (acquired in step <b>2701</b> as an x-ray projection image captured while x-ray source <b>104</b> is in the 0° scan angle position) is used to bound the shrink-wrapping algorithm being performed in step <b>2707.</b> Thus, an orthogonal projection image is extracted from the plurality of projection images acquired in step <b>2701.</b> Because, as described above, an orthogonal projection image is defined as the x-ray projection image captured while the x-ray source <b>104</b> is in the 0° scan angle position, no reconstruction is necessary or performed in step <b>2702</b> to extract the orthogonal projection image. Accordingly, in each slice, the shrink wrapping is bounded by an orthogonal image of the object(s) in the region-of-interest received in step <b>2705.</b>

0158In step <b>2709,</b> an outline of the imaged object (i.e., object <b>50</b>) is formed in each slice of the stack, based on the shrink-wrapping algorithm bounded by the orthogonal image. Thus, for each slice, an outline of the structure in the region-of-interest is obtained. In addition, in step <b>2709,</b> each contour corresponds to a single slice through the volume. The surface of that volume is smoothed to eliminate nonuniformities by, for example, Gaussian filtering of the volume or by Laplace flow smoothing of a surface mesh.

0159In step <b>2710,</b> there is a determination, based on integrated edge strength adjacent to the contour, of whether the last slice of the object within the image stack has been processed in the loop. More specifically, the gradient magnitude is integrated along the edge contour to estimate the overall edge strength. If not, the procedure returns to step <b>2706</b> to process a next slice in the stack. On the other hand, if the last slice has been processed, the procedure proceeds to step <b>2711.</b>

0160In step <b>2711,</b> the outlines formed in step <b>2709</b> are up-sampled in the z-dimension, and linear or trilinear interpolation is performed in-between the outlines to provide a higher-resolution contour. In particular, the collection of outlines, e.g., the set comprising each respective outline for each slice in the stack obtained for the region-of-interest in step <b>2709,</b> represents a discretized contour of a 3D surface, and interpolating between the outlines will provide a highly-sampled and smooth 3D contour.

0161In step <b>2712,</b> the nodes of neighboring outlines are aligned in the z-dimension. Any discontinuities in the volume are eliminated by aligning the final position of the grid points of the active contour of neighboring outlines. This may be done by linearly interpolating the grid positions between those estimated in adjacent slices.

0162In step <b>2713,</b> the 3D surface contour is created from the aligned outlines. In this manner, the output of this technique is a 3D rendering of the dataset which is subsequently displayed as shown, for example, as view <b>2904</b> in <figref idref="f0037">FIG. 29</figref> or as view <b>3004</b> in <figref idref="f0038">FIG. 30</figref>, which will be discussed more fully below.

0163In step <b>2714,</b> the 3D surface contour obtained in step <b>2713</b> is displayed or otherwise presented. The 3D contour can, for example, be displayed on a display screen in a separate window, as shown by view <b>2904</b> in <figref idref="f0037">FIG. 29</figref> or view <b>3004</b> in <figref idref="f0038">FIG. 30</figref>. The display may be provided with lighting, rotational, camera, coloring, viewing controls for the 3D contour, and/or the like. Thus, for example, the contour can be rotated freely and a full 3D visualization, including side and back views, of the structures can be provided. In addition, measurement capabilities can be incorporated to provide true 3D measurements between multiple structures/features, as described more fully below with respect to <figref idref="f0039 f0040 f0041">FIG. 31A to FIG. 31C</figref>.

0164<figref idref="f0036">FIG. 28</figref> is a flowchart (titled "Pre-Processing") illustrating a procedure for pre-processing reconstructed tomosynthesis slices according to an example embodiment herein, and represents in further detail the manner in which step <b>2703</b> in <figref idref="f0033">FIGS. 27A</figref> and <figref idref="f0034">27B</figref> is performed. In step <b>2801,</b> edge enhancement is performed for each slice reconstructed in step <b>2702,</b> for example, by use of Fourier filter which ramps linearly from 100% to 30% ranging from maximum spatial frequency to zero spatial frequency. Thus, in one example herein, a filter is used to increase the edge contrast of each structure in the slice in order to improve its sharpness.

0165In step <b>2802,</b> noise reduction filtering is performed on each of the slices, using, for example, a low-pass filter. Similarly, for example, each volume slice may be convolved with a Gaussian kernel with a radius between 0.25 and 1 pixel in order to reduce noise. Accordingly, noise and other artifacts can be removed or reduced from the slices.

0166In step <b>2803,</b> edge detection filtering is performed on each slice, to more clearly define the edges of each structure (e.g., object <b>50,</b> such as one or more teeth). Typically, edge detection filtering may be performed in many manners. One technique to estimate the image gradient in a given direction is by convolution with a kernel estimated from the derivative of a Gaussian in the given direction. As a result of the pre-processing described with respect to <figref idref="f0036">FIG. 28</figref>, the visibility of the slices is enhanced.

0167<figref idref="f0037">FIG. 29</figref> is a view for illustrating a display and user interface according to an example embodiment herein.

0168In particular, as shown in <figref idref="f0037">FIG. 29</figref>, multiple windows can be presented (via, e.g., display unit <b>108</b>) to allow for convenient control and analysis by a viewer.

0169In the example shown in <figref idref="f0037">FIG. 29</figref>, image <b>2901</b> is an image for defining a region-of-interest <b>2906,</b> as described above with respect to step <b>2704</b> of <figref idref="f0033">FIGS. 27A</figref> and <figref idref="f0034">27B</figref>. An example of a 3D rendering (of the region-of-interest <b>2906</b>) is shown in image <b>2904,</b> and can be obtained according to the procedure described in <figref idref="f0033">FIG. 27A</figref>.

0170As shown in <figref idref="f0037">FIG. 29</figref>, an image <b>2901</b> is an orthogonal projection image obtained in step <b>2701,</b> and corresponds to an image captured while the x-ray source <b>104</b> is in the 0° position shown in <figref idref="f0001">FIG. 1A</figref> above. The orthogonal image <b>2901</b> can be displayed alongside an image <b>2902</b> of a current slice (as shown), in order to provide a reference view for comparison. Thus, in one embodiment, the orthogonal projection image <b>2901</b> is displayed as a default image regardless of whether the user has selected another slice, and is displayed separately from any currently selected slice, and, so as to provide a visual default reference.

0171Image <b>2902</b> depicts a currently selected tomosynthesis slice in the stack. Specifically, image <b>2902</b> is a currently selected image from the tomosynthesis stack (which was reconstructed in step <b>2702</b> from the projection images obtained in step <b>2701</b>) that has undergone preprocessing in step <b>2703.</b> A user may change the displayed tomosynthesis slice <b>2902</b> by navigating to different slices in the stack. In particular, using an input unit (e.g., input unit <b>114</b>), the user can view images at different planes in the stack. For example, in the embodiment shown in <figref idref="f0037">FIG. 29</figref>, the user can click or otherwise designate arrow indicators on scroll bar <b>2905</b> to toggle through each slice to navigate to a desired slice. Scroll bar <b>2905</b> may also include additional visual indicators of the position of the currently displayed slice in the stack. For example, in the embodiment shown in <figref idref="f0037">FIG. 29</figref>, scroll bar <b>2905</b> displays the slice position relative to the detector plane for the corresponding slice image <b>2902.</b> The scroll bar may, for example, indicate whether the slice is towards a buccal or lingual position.

0172<figref idref="f0038">FIG. 30</figref> is a view for illustrating another example display and user interface according to an example embodiment herein. In <figref idref="f0038">FIG. 30</figref>, the orientation and function of images correspond generally to those depicted in <figref idref="f0012">FIG. 6</figref>, but after a rotation of the 3D rendering (see <b>3004</b>). Thus, in <figref idref="f0038">FIG. 30</figref>, image <b>3001</b> corresponds to an image of the slice with the region-of-interest remaining highlighted, image <b>3002</b> depicts the current slice, image <b>3003</b> depicts the current position of the sensor, and scroll bar <b>3005</b> is for scrolling through the slices. Image <b>3004</b> depicts the 3D rendering of the tooth and may be rotated (in any direction) to allow additional 3D viewing of the structure of the object of interest (e.g. a root). The example of a 3D rendering shown in image <b>3004</b> can be obtained according to the procedure described in <figref idref="f0034">FIG. 27B</figref>. In one example embodiment, display in step <b>2714</b> is updated according to user input. In particular, the user can manipulate the 3D image in view <b>3004</b> to orient the displayed object in the manner desired, by, for example, accessing viewing controls or predefined input device actions (e.g., via input unit <b>114</b> shown in <figref idref="f0001">FIG. 1A</figref>), to thereby view different aspects of the tooth (or other structure). Accordingly, the 3D image is displayed in step <b>2714</b> (e.g., via display unit <b>108</b> shown in <figref idref="f0001">FIG. 1A</figref>), and visual aspects of the 3D image can be manipulated to provide a different display.

0173<figref idref="f0039 f0040 f0041">FIGS. 31A to 31C</figref> are views for illustrating measurement of distances in a 3D image according to example embodiments herein.

0174Specifically, <figref idref="f0039 f0040 f0041">FIGS. 31A to 31C</figref> show different states of a 3D rendering image of a tooth depicted on a display unit (e.g., via display unit <b>108</b> shown in <figref idref="f0001">FIG. 1A</figref>), as a user sets measurement points and enables measurement of distances on the 3D rendering.

0175Specifically, <figref idref="f0039">FIG. 31A</figref> is an illustration representing examples of how a measurement process between two points on a 3D rendering of a tooth can be started. As shown in image <b>3101,</b> a point on the 3D rendering of a tooth (e.g., using a mouse or other interface device) is navigated to.

0176In image <b>3102,</b> with the desired first measure point "A" having been navigated to, a user manipulates an input device (e.g., via input unit <b>114</b> shown in <figref idref="f0001">FIG. 1A</figref>) to display a user interface (e.g., menu <b>3103),</b> for example by right-clicking on a mouse. Menu <b>3103</b> enables a user to, for example, rotate or manipulate the 3D rendering, set mouse controls, set hot keys, obtain further information ("about"), and measure between points on the 3D rendering. In the example shown in image <b>3102,</b> the user has selected the "measure" option <b>3104,</b> which brings up a further user interface <b>3105</b> enabling the user to set a measure point A, a measure point B, to request the distance between A and B, and to reset. As shown, the user selects "Set Point A" to set a first measurement point (indicated by the triangle in <figref idref="f0039">FIG. 31A</figref>).

0177Turning to <figref idref="f0040">FIG. 31B</figref>, the user follows a similar procedure to set the measure point B. As can be seen from image <b>3106,</b> however, the 3D rendering has been rotated from the views shown in <figref idref="f0039">FIG. 31A</figref>. Thus, image <b>3106</b> represents an example of a case where the user has rotated and manipulated the view for better access to measure points desired for selection. Once measure point B is selected (represented by the cross in <figref idref="f0040">FIG. 31B</figref>), image <b>3107</b> is provided with a line <b>3108</b> connecting the two points A and B.

0178In <figref idref="f0041">FIG. 31C</figref>, in image <b>3109,</b> the user selects to calculate the distance between the points A and B that were selected as shown in <figref idref="f0039">FIG. 31A</figref> and <figref idref="f0040">FIG. 31B</figref>. This can be accomplished with, for example, the same drop-down menu used to set the measurement points A and B. As a result, the distance between points A and B is calculated. For example, assuming the coordinates of point A in the 3D space are (x₁, y₁, z₁) and the coordinates of point B in the 3D space are (x₂, y₂, z₂), the distance can be calculated as d=√{square root over ((x₂-x₁)₂+(y₂-y₁)²+(z₂-z₁)²)}.

0179In image <b>3110,</b> a window <b>3111</b> appears, displaying the calculated distance between points A and B, along with coordinates of the points A and B in the 3D space. Accordingly, the user is provided with a convenient tool to easily view and measure distances between structures (or distances within structures) in the 3D rendering, whether the structures are in the same image slice or different image slices.

0180As discussed, for x-ray images to have value and utility in clinical diagnosis and treatment, they should have high image fidelity and quality (as measured by resolution, brightness, contrast, signal-to-noise ratio, and the like, although these example metrics are not limiting) so that anatomies of interest can be clearly identified, analyzed (e.g., analysis of shape, composition, disease progression, etc.), and distinguished from other surrounding anatomies. The processes described in <figref idref="f0033">FIGS. 27A</figref> and <figref idref="f0034">27B</figref> can be used to improve image fidelity and quality by reducing image artifacts and optimizing image contrast.

Measuring Three-Dimensional Distances in a Stack of Images

0181In accordance with an example aspect described herein, a method, system, apparatus, and computer program product are provided for measurement between structures in tomosynthesis images, and more particularly, for measuring 3D distances in a stack of tomosynthesis images will now be further described in conjunction with <figref idref="f0042">FIG. 32</figref>, which shows a flow diagram of an exemplary process for measuring distances in a three-dimensional tomosynthesis images according to one example embodiment herein.

0182Briefly, by virtue of the procedure of <figref idref="f0042">FIG. 32</figref> and as will be described below, distances between points in a three-dimensional space corresponding to a tomosynthesis dataset may be obtained. A stack of tomosynthesis images (slices) is obtained. Specifically, the stack of tomosynthesis images is reconstructed from x-ray projection images taken over a scan angle. The tomosynthesis slices are a series of images which are parallel to a plane defined by the surface of the x-ray detector <b>102.</b> The tomosynthesis stack is comprised of the tomosynthesis slices. Each tomosynthesis slice is stacked on top of or below another slice in a direction orthogonal to the plane of the x-ray detector. Thus, the plurality of tomosynthesis slices comprising the tomosynthesis stack correspond to an image volume (i.e., a certain volume of space in three-dimensions). If the plane of the x-ray detector is defined as the x-y plane and a direction orthogonal to the plane of the x-ray detector is defined as the z-axis, then a point P_i within the image volume may be identified by it three-dimensional Cartesian coordinates (x_i, y_i, z_i).

0183In the exemplary embodiment shown in <figref idref="f0042">FIG. 32</figref>, a distance from one point in the image volume to another point in the image volume may be determined. Briefly, a first point on a first tomosynthesis slice and a second point on a second tomosynthesis slice (or the first tomosynthesis slice) are selected. The computer system <b>106</b> may determine the x-y coordinates of the first and second points from the pixel arrays corresponding to the respective tomosynthesis slices. As noted above, each tomosynthesis slice occupies a certain position within the tomosynthesis stack. In other words each tomosynthesis slice lies a certain distance from the surface of the x-ray detector along the z-axis. Thus, all the points within a tomosynthesis slice have the same value for a z-coordinate. Thus, by knowing the system geometry and the position of the tomosynthesis slice within the tomosynthesis stack, the z-coordinate may be determined.

0184As one of ordinary skill will appreciate, the x-y coordinates of a given point may correspond to a particular anatomical plane. For example, a tomosynthesis slice may lie in a plane substantially parallel to the occlusal surface. The relationship between the tomosynthesis slices and certain anatomical features will be further explained below in reference to <figref idref="f0049 f0050 f0051">FIGS. 39A-41B</figref>.

0185In step <b>3200,</b> the computer system <b>106</b> generates a tomosynthesis stack comprising a plurality of tomosynthesis slices from a plurality of projections images, as discussed above and with reference to <figref idref="f0001 f0002 f0003 f0004">FIGS. 1A-1D</figref>.

0186In step <b>3201,</b> a user selects a tomosynthesis slice from the tomosynthesis stack. In an exemplary embodiment, the user may operate an input unit (e.g., input unit <b>114</b>) to "scroll" or "page" through each, or selected ones, of the slices in the tomosynthesis stack, while a display unit (e.g., display unit <b>108</b>) displays those slices. To select a particular slice, the user may stop scrolling or otherwise moving through the stack. An example embodiment illustrating this process is shown in <figref idref="f0043">FIGS. 33A-33D</figref>, which will be discussed more fully below.

0187Once the user has selected a particular slice, the user may, in step <b>3202,</b> use the input unit <b>114</b> to select a point on the displayed slice to place a marker, which then serves a first measurement point. In an exemplary embodiment, this marker indicates a particular anatomical location or plane. The user may also change the location of the marker after it is initially set. For example, the user may manipulate the input unit <b>114</b> (e.g., a mouse) to move the marker over the displayed slice, and then use another input (e.g., clicking a button on the mouse) to designate the placement of the marker once the marker is in a desired location (step <b>3203</b>).

0188Once the marker is placed at the desired location, the location of the marker is set as a point A, and the x-y coordinates of point A are saved in memory as point A (or using some other designation). The marker for point A may be displayed differently from a marker in another later selected slice for a point B, in order to help the viewer quickly observe that the points are in different slices and thus at different positions along the z-axis. For example, a marker for point A may be displayed in a different color than a marker for point B. An example embodiment illustrating this process is shown in FIGS. 34A-<figref idref="f0047">37D</figref>, which will be discussed more fully below.

0189Turning now to step <b>3207,</b> before or concurrently with the actions in steps <b>3201</b> to <b>3203,</b> a three-dimensional volumetric image ("volumetric image") of the image volume is created. The volumetric image may provide for better visualization and representation of an imaged object. For example, if the imaged object is a tooth, the volumetric image may provide a user with a perspective view of the tooth which aids in determining a relative location of the slices. The volumetric image of the image volume may be generated from the tomosynthesis stack. As discussed above, the tomosynthesis stack includes a plurality of tomosynthesis slices at a plurality of depths, and the volumetric image of the image volume may be generated from the tomosynthesis slices. In another example embodiment herein, the volumetric image may represent the three-dimensional structure of the imaged anatomy or it may represent a model volume that approximates the shape of the anatomy.

0190As discussed above, each tomosynthesis slice in the tomosynthesis stack corresponds to a two-dimensional image in an x-y plane (parallel to the plane of the x-ray detector). The z-axis is orthogonal to the surface of the x-ray detector <b>102</b> and thus a coordinate value for the z-axis represents a distance from the surface of the x-ray detector <b>102.</b> Thus, as discussed above, scrolling through the tomosynthesis slices, corresponds to traveling through the image volume in a direction towards or away from the surface of the x-ray detector.

0191If the system geometry is known (or deduced from one or more objects within the plurality of projection images such as alignment markers) the computer system <b>106</b> may determine a distance between two tomosynthesis slices. For example, if a tomosynthesis slice lies just beyond the surface of a tooth (e.g., proximate to but not including, for example, a buccal surface), the computer system <b>106</b> may identify that slice and produce a depth measurement based on the separation between that slice and another slice. For example, if a slice proximate to but not including a buccal surface of a tooth is labelled S_i, and another slice located beneath, and within the tooth structure is labelled S_j, then the depth of slice S_j relative to the buccal surface may be calculated based on the difference between i and j and the known distance between slices.

0192In step <b>3208,</b> the marker placed at point A is transferred to a corresponding location on the volumetric image. The coordinates of point A may also be set as a geometrical base (i.e., an origin) of a measurement vector from point A to point B (discussed below), in the image volume.

0193To determine point B (step <b>3205</b>), a user may scroll through the tomosynthesis stack to a desired tomosynthesis slice (step <b>3204</b>) and then place a second marker (step <b>3206</b>) at a desired location (point B) on the displayed tomosynthesis slice. The x-y coordinates of point B are saved in memory under that designation (or some other designation). As discussed above, a user may operate the input unit (e.g., input unit <b>114</b>) to scroll or page through each of the slices in the tomosynthesis stack, and the display unit (e.g., display unit <b>108</b>) displays each slice. As discussed, a user may (at least temporarily) pause or stop scrolling, toggling, or otherwise moving through the stack when the user reaches the desired slice. Typically, the second marker would be placed on a different tomosynthesis slice (S₂) from the tomosynthesis slice containing the first marker. If the second marker were placed on the same slice as the first marker, then the measurement would correspond to a linear distance between the first and second marker within a plane defined by the tomosynthesis slice. In other words, a two-dimensional measurement as opposed to a three-dimensional measurement. Nevertheless, it should be understood that the second marker, and thus the second measurement point, could be in the same tomosynthesis slice as that of point A.

0194The second marker may be visually distinguishable from the first marker. For example, the second marker may be a different color, shape, or size from the first marker. Moreover, the first marker may remain visible on the displayed tomosynthesis slice, even if the displayed tomosynthesis slice is different from the tomosynthesis slice in which the first marker is placed. The user may therefore see the x-y location of the first marker as the user scrolls through the tomosynthesis stack.

0195As the user operates the input unit <b>114</b> to place the marker for the second measurement point, the computer system <b>106</b> calculates the measurement vector from point A to point B, which respectively correspond to the first and second measurement points in the image volume (step <b>3209</b>). If the user moves the second measurement point (or the first measurement point) to a different location using the input unit <b>114,</b> the computer system <b>106</b> may dynamically update a the measurement vector as the measurement point is moved. Thus, in one embodiment, the computer system <b>106</b> may dynamically update the measurement vector while either the first or second measurement point is moved or changed, in addition to calculating the measurement vector when the two measurement points are initially set.

0196In a similar manner to step <b>3208,</b> a point corresponding to point B is placed on the volumetric image of the image volume generated in step <b>3207</b> (step <b>3210</b>). As discussed, the volumetric image provides the user with a three dimensional view of the image volume. Thus, by placing points corresponding to points A and B, respectively, on the volumetric image, the user may appreciate the three dimensional relationship between points A and B, which may be not be obvious from the two-dimensional tomosynthesis slices.

0197In step <b>3211,</b> the computer system <b>106</b> calculates the vector V_AB between points A and B. The vector includes both a magnitude and direction. For example, assuming the coordinates of point A in the 3D space are (x₁, y₁, z₁) and the coordinates of point B in the 3D space are (x₂, y₂, z₂), the vector magnitude may be calculated according to d=√{square root over ((x₂-x₁)²+(y₂-y₁)²+(z₂-z₁)²)}. The computer system <b>106</b> may also calculate the x, y, and z vector components. These vector components may also correspond to the mesiodistal, buccolingual and coronoapical axes as discussed below.

0198Of course, other methods or formulas for calculating the magnitude of the vector can be used in step <b>3211.</b> The magnitude of the vector can then be displayed on the display unit <b>108</b> in a variety of manners, as discussed more fully below. The features described above and illustrated in <figref idref="f0042">FIG. 32</figref> will now be explained in further detail below with reference to <figref idref="f0043 f0044 f0045 f0046 f0047 f0048 f0049 f0050 f0051">FIGS. 33A-41B</figref>.

0199<figref idref="f0043">FIG. 33A</figref> is a two-dimensional projection image <b>3301</b> recorded by the x-ray detector <b>102</b> at the 0° position, as shown in <figref idref="f0001">FIG. 1A</figref> above. As discussed above, the plurality of projection images are used by computer system <b>106</b> to create a tomosynthesis stack comprising a plurality of tomosynthesis slices. <figref idref="f0043">FIG. 33B</figref> illustrates the spatial relationship between a tomosynthesis slice <b>3302</b> (shown in <figref idref="f0043">FIG. 33C</figref>) located a distance d₁ from the x-ray detector <b>102</b> in the z-axis direction. The tomosynthesis slice is number S₁ of S_i comprising the tomosynthesis stack, as indicated by slice number display <b>3304</b> adjacent to scroll-bar <b>3306.</b><figref idref="f0043">FIG. 33D</figref> is an illustration showing the location of the tomosynthesis slice <b>3302</b> within the image volume.

0200In an exemplary embodiment, one or more of the images shown in <figref idref="f0043">FIGS. 33A-33D</figref> may be displayed at the same time on a display unit (e.g., display unit <b>108</b>), so that a user can conveniently view information relating to a current slice and its relative position.

0201As discussed above, a user may use an input unit <b>114</b> to change the displayed tomosynthesis slice. For example, the user may interact with scroll-bar <b>3306</b> to change the displayed tomosynthesis slice. The user may drag the scroll-bar icon positioned on scroll bar <b>3306</b> to toggle through each slice to navigate to a desired slice in the z-axis direction. Scroll bar <b>3306</b> may also include additional visual indicators of the position of the current slice in the stack. For example, a text box may be displayed concurrently with the tomosynthesis slice that shows a distance (di) from the tomosynthesis slice to the surface of the x-ray detector <b>102.</b> As the user scrolls through the tomosynthesis stack, the value d_i may be concurrently updated to correspond to the displayed tomosynthesis image.

0202As discussed above in regard to steps <b>3202</b> and <b>3203,</b> once a user selects a desired tomosynthesis slice, a marker <b>3402</b> may be placed at a desired location, as illustrated in <figref idref="f0044">FIG. 34</figref>. In one embodiment, the user may place the marker <b>3402</b> by a double-tap operation of a mouse, touchscreen, or similar device. In another embodiment, the user may use the input <b>114</b> to bring up a menu graphical user interface menu ("menu") that may include a command "Set As Point A," along with other commands (e.g., "Set As Point B", "Clear Point A", "Clear Point B", "Clear All Points", "Rotate Measurement Axis", "Reset 3D View"). Some of the commands may be "greyed out", i.e., unavailable, during the process illustrated in <figref idref="f0042">FIG. 32</figref>. By selecting the command "Set As Point A", the desired location may be set as point A, and a marker <b>3402</b> may be placed at that location.

0203As discussed above in reference to steps <b>3204-3206,</b> once point A is selected, the user may select another point (point B) on the same tomosynthesis slice or a different tomosynthesis slice. As discussed above, the user may use input unit <b>114</b> to navigate to the desired tomosynthesis slice (if applicable). <figref idref="f0045">FIG. 35C</figref> is an illustration of a second tomosynthesis slice <b>3502</b> (S₂) located a distance d₂ from the x-ray detector <b>102</b> (as illustrated in FIGS. 6B and 6D) which the user has navigated to. As shown in <figref idref="f0045">FIG. 35C</figref>, the slice number display <b>3304</b> now reads S₂. <figref idref="f0045">FIG. 35A</figref> is the x-ray projection image recorded at the 0° position and is identical to <figref idref="f0043">FIG. 33A</figref>, discussed above. As discussed above, one or more of these images may be displayed on display unit <b>108,</b> either concurrently or individually. While <figref idref="f0045">FIGS. 35B-35D</figref> illustrate a tomosynthesis slice <b>3502</b> which is further away from the surface of the x-ray detector <b>102</b> in the z-axis direction, the second tomosynthesis slice could also be located closer to the x-ray detector <b>102.</b> As discussed, a text box may also be displayed on the display unit <b>108</b> which indicates the distance from the selected tomosynthesis slice <b>3502</b> to the x-ray detector <b>102.</b> It should be noted, that the distance from the tomosynthesis slice <b>3502</b> to the x-ray detector <b>102</b> is only one type of distance that could be displayed. If, as discussed above, the computer system <b>106</b> determines that a particular tomosynthesis slice corresponds to a position proximate to, but not including, a surface of the imaged object, a distance from that tomosynthesis slice to any other tomosynthesis slice may also be calculated and displayed (either alone or along with another distance). Such a distance calculation may provide an approximate depth from the surface of the imaged object to the displayed tomosynthesis slice.

0204As shown in <figref idref="f0045">FIG. 35C</figref>, the marker <b>3402</b> placed at point A may be visible in the displayed tomosynthesis slice <b>3502.</b> In an exemplary embodiment, the marker <b>3402</b> may be displayed differently when a different tomosynthesis slice is displayed. For example, as shown in <figref idref="f0045">FIG. 35C</figref>, marker <b>3402</b> is filled with dashed lines which indicate that the marker <b>3402</b> was placed in a different tomosynthesis slice from the one displayed (tomosynthesis slice <b>3502</b>). Alternatively, one or more of the size, shape, color, fill pattern, and transparency of marker <b>3402</b> may also be changed.

0205<figref idref="f0046">FIG. 36</figref> is an enlarged view of tomosynthesis slice <b>3502</b> shown in FIG. 6C. As discussed above, a user may place a second marker <b>3602</b> at a desired location. In the example shown in <figref idref="f0046">FIG. 36</figref>, the second marker <b>3602</b> is a triangular shape to distinguish it from the first marker <b>3302.</b> However, the markers <b>3402</b> and <b>3602</b> may be the same shape, but have different colors or sizes to distinguish one from the other. The markers <b>3402</b> and <b>3602</b> may also have different fill patterns or transparencies to distinguish one from the other. As discussed, the second marker <b>3602</b> may be set through an input from the input unit <b>114.</b> For example, if the user is using a mouse, the user may right-click to bring up a menu (as discussed above) that includes one or more commands, such as "Set As Point A", "Set As Point B", "Clear Point A", "Clear Point B", "Clear All Points", "Rotate Measurement Axis", "Reset 3D View".

0206As shown in <figref idref="f0047">FIGS. 37A-37D</figref>, once the second marker <b>3602</b> is set, the vector (V_ab) <b>3702</b> may be calculated from the first marker <b>3402</b> to the second marker <b>3602</b> and displayed three-dimensionally in a coordinate system representing the image volume (see <figref idref="f0047">FIG. 37D</figref>). As discussed above, the vector magnitude <b>3704</b> (D_ab) may be calculated and displayed as well.

0207In an exemplary embodiment, however, the user does not have to set the second marker <b>3602</b> in order to calculate and display a corresponding vector. Rather, as the user moves the second marker <b>3602</b> over the displayed tomosynthesis slice <b>3502,</b> the computer system <b>106</b> dynamically tracks the movement of the second marker <b>3602</b> and produces the vector magnitude <b>3704</b> and the displayed vector <b>3702</b> between points A and B on the volumetric view of the image volume (see <figref idref="f0047">FIG. 37D</figref>). As discussed in further detail below, depending upon the type of diagnostic image and the selected points, the displayed vector <b>3702</b> may correspond to certain anatomical measurements, such as, for example a mesiodistal length, coronoapical length, and buccolingual length

0208Accordingly, even without setting the second marker <b>3602</b> at the second measurement point (point B), the user is provided with the vector magnitude and a visual depiction of the vector from the first measurement point (point A) represented by the first marker <b>3402</b> to the second measurement point (point B) represented by the second marker <b>3602.</b> As such, the vector is dynamically updated as the second marker is moved to a new location.

0209In an exemplary embodiment, the volumetric view of the image volume may be rotated to provide the user with a different viewing angle, as illustrated in <figref idref="f0048">FIGS. 38A and 38B. FIG. 38A</figref> is identical to <figref idref="f0047">FIG. 37D</figref>, which is a default viewing angle for the volumetric view. <figref idref="f0048">FIG. 38B</figref> is another volumetric view of the image volume, which has been rotated to provide an additional perspective view. The particular perspective view shown in <figref idref="f0048">FIG. 38B</figref> is only exemplary. The image volume (and the calculated vectors displayed therein) may be viewed from any angle or perspective. In addition, the user may select from a GUI menu one or more preset views, such as, for example, an X-Y view, an X-Z view, and a Y-Z view. Therefore, according to the example embodiment herein, a view of the three-dimensional coordinate space may be rotated.

0210In an exemplary embodiment, a user may quickly navigate through the tomosynthesis stack using one or more of the markers. As discussed, the computer system <b>106</b> may cause the display unit <b>108</b> to change the displayed tomosynthesis slice based on a command received through the input unit <b>114.</b> If a user selects a marker corresponding to a tomosynthesis slice which is not currently displayed, the user may enter a command through the input unit <b>114</b> which will cause the computer system <b>106</b> to retrieve and display the corresponding tomosynthesis slice. The user may select the marker by, for example, double-clicking on the marker, or by right-clicking on the marker to bring up GUI menu system, and then selecting an appropriate command, such "Display Corresponding Tomosynthesis Slice". For example, if tomosynthesis slice <b>3502</b> is currently displayed (as shown in <figref idref="f0047">FIG. 37C</figref>), the user may double-click on the first marker <b>3402</b> (corresponding to tomosynthesis slice <b>3302</b>), and the computer system <b>106</b> will then cause the display unit <b>108</b> to display tomosynthesis slice <b>3302</b> (see <figref idref="f0043">FIG. 33C</figref>). The slice number display <b>3304</b> will then display the corresponding tomosynthesis slice number (e.g., S₁). As such, it is possible for a user to quickly return to a view where a marker was originally set, or other views where other markers were set.

0211While the above description has detailed the placement of first and second markers, more markers may be placed, either on the same tomosynthesis slice or other tomosynthesis slices. The computer system <b>106</b> may, in a manner similar to that described above, calculate and display additional vectors corresponding to these markers on display unit <b>108.</b> The respective magnitudes of these vectors may also be displayed (like in <figref idref="f0047">FIG. 37D</figref>), either concurrently with other vector magnitudes or individually.

0212<figref idref="f0049 f0050 f0051">FIGS. 39A-41B</figref> illustrate another exemplary embodiment. <figref idref="f0049">FIGS. 39A-39C</figref> are illustrative of three images which may be simultaneously or individually displayed on display unit <b>108.</b> In <figref idref="f0049">FIG. 39A</figref>, a two-dimensional x-ray projection image <b>3902</b> corresponding to the 0° angle shown in <figref idref="f0001">FIG. 1A</figref> is displayed on display unit <b>108.</b> In the manner discussed above, a user may select a tomosynthesis slice <b>3904</b> from the tomosynthesis stack (using, for example, scroll-bar <b>3306</b>), which may also be displayed on display unit <b>108.</b> In that regard, the user may use a scroll bar <b>3306</b> to scroll between tomosynthesis slices in the tomosynthesis stack, and the corresponding slice number may be displayed in slice number display <b>3304.</b> As discussed above, the computer system <b>106</b> may generate a three-dimensional volumetric image <b>3906</b> (<figref idref="f0049">FIG. 39C</figref>) of the image volume. In the exemplary embodiment illustrated in <figref idref="f0049">FIG. 39C</figref>, the three-dimensional volumetric image <b>3906</b> is displayed on display unit <b>108</b> as opposed to a volumetric view of the image volume (e.g., <figref idref="f0043">FIG. 33D</figref>). However, both the three-dimensional volumetric image <b>3906</b> and a volumetric view of the image volume may be displayed on display unit <b>108</b> at the same time.

0213As illustrated in <figref idref="f0049">FIG. 39C</figref>, coordinate axes <b>3908</b> may also be provided to show the user the viewing orientation of the volumetric image. Rather than labeling the coordinate axes in Cartesian nomenclature, the axes may be described using anatomical nomenclature (e.g., mesiodistal (M), coronoapical (C), and buccolingual (L)) to provide the user with information regarding the anatomical orientation of the image volume. In an exemplary embodiment, the ends of the scroll bar <b>3306</b> may be labelled with corresponding descriptions of the axes. For example, if the particular diagnostic image results in the tomosynthesis stack being arranged along the buccolingual axis then the ends of the scroll bar <b>3306</b> may labelled "buccal" and "lingual". Thus, the user can easily determine which direction to scroll the slider to move in one anatomical direction or another.

0214As described above, once the user has selected a particular tomosynthesis slice, a first marker <b>3910</b> may be placed at a first measurement point (point A), as illustrated in <figref idref="f0049">FIG. 39B</figref>. The user may then select a second tomosynthesis slice <b>4002</b> and place a second marker <b>4004</b> to designate a second measurement point (point B), as illustrated in <figref idref="f0050">FIG. 40B</figref>. As described above, the second tomosynthesis slice <b>4002</b> may be the same slice or a different slice. As discussed, the user may move the second marker <b>4004</b> to any location on the second tomosynthesis slice <b>4002</b> and the computer system <b>106</b> may dynamically calculate and update the magnitude of a vector <b>4006</b> from the first measurement point to the second measurement point. The magnitude of that vector (corresponding to a distance between the two points) may be displayed, along with the magnitudes of the individual vector components, in a text box <b>4008</b> (as illustrated in <figref idref="f0050">FIG. 40C</figref>). As shown in <figref idref="f0050">FIG. 40C</figref>, the vector <b>4006</b> may be displayed in partial relief, that is the vector <b>4006</b> may be partially shaded when the vector traverses through a structure. The vector <b>4006</b> may also be shown in relief by other visual means such as, for example, a dashed line.

0215In an exemplary embodiment, the volumetric image <b>3906</b> may be rotated by the user to provide different views of both the volumetric image <b>3906</b> and the vector <b>4006</b> contained therein. For example, as shown in <figref idref="f0051">FIG. 41A</figref>, the volumetric image <b>4006</b> has been rotated to provide a clearer view of the occlusal surface of the imaged tooth. The coordinate axes <b>3908</b> have also been rotated to provide the user with an indication of the anatomical directions. In <figref idref="f0051">FIG. 41B</figref>, the volumetric image <b>3906</b> has been rotated to provide a view of the coronoapical-mesiodistal plane (i.e., in a direction of the buccolingual axis). For simplicity, vector <b>4006</b> has not been shown in relief in <figref idref="f0051">FIGS. 41A-41B</figref>, however, the computer system <b>106</b> may cause the display unit <b>108</b> to display vector <b>4006</b> in relief to provide additional visualization of the vector <b>4006</b> for the user.

0216The volumetric image <b>3906</b> may be rotated through a variety of means. For example if the input unit <b>114</b> is a mouse, the user may click on a part of the display volumetric image <b>3906</b> and drag the mouse to displayed volumetric image <b>3906</b> to rotate in a corresponding manner. In another example, a GUI menu may be displayed which shows one or more icons that may be clicked on by the user to cause the volumetric image to rotate. In yet another example, the computer system <b>106</b> may display a list of predetermined viewing perspectives which the user may select from. If the type of diagnostic image is known, a set of corresponding views which may be useful for the particular diagnostic operation may be presented in the form a GUI menu by the computer system <b>106.</b> The user may then select one of those views which will be displayed by display unit <b>108.</b> For example, if the diagnostic operation is a measurement of thickness of the lingual and buccal plates, a view along the mesiodistal axis (i.e., of the buccolingual plane) may be predetermined view selectable from the GUI menu.

0217Accordingly, by virtue of the example embodiments described herein, 3D measurement in a stack of tomosynthesis images is provided, even across or within images of the stack. For example, it is possible to measure the distances between points on objects or structures represented by tomosynthesis images, even though the images themselves are only two-dimensional. Moreover, it is possible to provide useful information based on such measurement.

Measuring and Visualizing Lingual and Buccal Plate Thickness

0218As described below, methods, systems, apparatuses, and computer program products are provided for measuring the thicknesses of the lingual and buccal plates.

0219<figref idref="f0052">FIG. 42A</figref> is a perspective view of the human mandible <b>4200.</b> The mandible <b>4200</b> includes the alveolar process <b>4202</b> (indicated as region above line A), which is a thickened region that contains the tooth sockets. As shown in <figref idref="f0052">FIG. 42B</figref>, the alveolar process <b>4202</b> includes a lingual bone plate <b>4204</b> and a buccal bone plate <b>4206</b> disposed on opposite sides of a tooth cavity for tooth <b>4212.</b> If a problem arises with tooth <b>4212</b> such that it must be replaced with an implant, the lingual plate <b>4204</b> and the buccal plate <b>4206</b> must have sufficient strength to support the implant. If the lingual plate <b>4204</b> and the buccal plate <b>4206</b> lack sufficient strength, a fracture may occur when the implant is exposed to forces incurred during the chewing process. The respective strengths of the lingual plate <b>4204</b> and the buccal plate <b>4206</b> are directly related to their respective thicknesses in a region that supports the implant. Traditional x-ray imaging in the periapical direction is incapable of resolving the thicknesses of the lingual and buccal plates because, for one, the imaging direction (i.e., the direction in which an x-ray travels from the x-ray source) is generally perpendicular to the lingual and buccal plates. However, the imaging direction is not the only reason traditional x-ray imaging is incapable of resolving the thicknesses of the lingual and buccal plates, as explained below.

0220<figref idref="f0053">FIGS. 43A and 43B</figref> are illustrations of the maxilla (<figref idref="f0053">FIG. 43A</figref>) and mandible (<figref idref="f0053">FIG. 43B</figref>) from the occlusal direction with the lingual plates <b>4204</b> and <b>4208</b> and the buccal plates <b>4206</b> and <b>4210</b> shown. Unlike when imaging in the periapical direction, here the lingual and buccal plates are substantially parallel to the imaging direction. However, as discussed above in reference to <figref idref="f0070">FIGS. 60A &amp; 60C</figref> an traditional x-ray image lacks depth information and overlapping objects easily obscure one another. For example, as shown in <figref idref="f0070">FIG. 60B</figref>, at least a portion of an incisor tooth <b>6008</b>A and a portion of the buccal plate <b>4206</b> lie in the same x-ray path. Thus, the incisor tooth <b>6012</b>A obscures at least a portion of the buccal plate <b>4206,</b> making it difficult to precisely determine the boundaries of the buccal plate <b>4206.</b> In a similar fashion tooth <b>6008</b>B obscures the view of the lingual plate <b>4204.</b>

0221Moreover, a thickness measurement of the lingual and buccal bones from occlusal direction using a traditional x-ray image may lead to an inaccurate measurement, as discussed below.

0222<figref idref="f0070">FIG. 60D</figref> shows cross-sectional views of the lingual plate <b>4204</b> and buccal plate <b>4206,</b> and the resulting x-ray projection images (<b>6012</b> and <b>6014</b>) from those plates for x-rays incident in the direction of arrow C. If one were to measure the width of the white portion of images <b>6012</b> and <b>6014</b> in an attempt to determine the thicknesses of the lingual plate <b>4204</b> and the buccal plate <b>4206</b> (as indicated by <b>6016</b> and <b>6018</b>) at a particular depth in the imaging direction (defined by ray C), the resulting measurement may over represent the thicknesses of the plates. This because the white portions in the image represent the widest points of the imaged object which may not be at the same depth. This is illustrated in <figref idref="f0070">FIG. 60D</figref>. The thickness measurement <b>6016</b> corresponds to a plane <b>6020</b> that is angled relative to the occlusal surface of the tooth (which is generally perpendicular to the imaging direction). Thus, the lack of depth information may lead to an improper diagnostic measurement.

0223<figref idref="f0054">FIG. 44</figref> is a flow chart illustrating the steps in measuring the thickness of the lingual and buccal plates, in accordance with one exemplary embodiment.

0224First, in step S<b>4402</b> the patient is positioned relative to the x-ray source <b>104.</b> More specifically, an x-ray detector <b>102</b> (e.g., one of the intraoral sensors described above) is carefully positioned inside of a patient's mouth. The patient bites down gently on the x-ray detector <b>102</b> to fix its position within the patient's mouth. A protective cover may be placed over the x-ray detector <b>102</b> to prevent the x-ray detector <b>102</b> from being damaged, as well as for sanitary reasons. Next, the x-ray source <b>104</b> is moved to an appropriate starting position for the radiographic scan (S<b>4404</b>). <figref idref="f0055">FIG. 45</figref> is an exemplary illustration of an x-ray source <b>104</b> mounted in such a manner that it may be positioned at any location within a three-dimensional space.

0225As shown in <figref idref="f0055">FIG. 45</figref>, the x-ray source <b>104</b> may be connected to an adjustable arm <b>4510,</b> which may be segmented and include one or more joints such as: a hinge, a swivel, a universal joint, or the like. The adjustable arm <b>4510</b> allows the x-ray source <b>104</b> to freely translate in three-dimensional space. Attached to one end of the adjustable arm <b>4510</b> is a vertical member <b>4520.</b> The other end of the adjustable arm <b>4510</b> may be mounted to a stationary structure, such as a wall or a ceiling. The vertical member <b>4520</b> is suspended vertically from the adjustable arm <b>4510</b> by a joint that allows the vertical member <b>4520</b> to freely rotate about an axis (A<b>1</b>) substantially defined by the vertical member <b>4520,</b> regardless of the position and orientation of the adjustable arm <b>4510.</b> The vertical member <b>4520</b> includes a bearing assembly which acts as a channel through the vertical member <b>4520.</b> A yoke <b>4530</b> is movably constrained within the channel, and can be angularly displaced through the bearing assembly and thus through the vertical member <b>4520.</b> A brake may hold the yoke <b>4530</b> in place and substantially prevent any motion of the yoke <b>4530</b> through the bearing assembly, thus locking the position of the yoke <b>4530</b> relative to the vertical member <b>4520.</b> A brake release button may also be provided such that an operator can release the brake and allow the yoke <b>4530</b> to rotate through the vertical member <b>4520.</b>

0226The motorized stage <b>118</b> may include arms <b>4560</b> and <b>4570</b> which are movably attached to the yoke ends <b>4540</b> and <b>4550,</b> respectively, each point of attachment forming a pivot such that the motorized stage <b>118</b> can be pitched about an axis (A<b>2</b>) which is substantially defined by the yoke ends <b>4540</b> and <b>4550</b> and substantially orthogonal to the axis (A<b>3</b>) of the x-ray source <b>104.</b> In the exemplary arrangement illustrated in <figref idref="f0055">FIG. 45</figref>, the x-ray source <b>104</b> may be appropriately positioned at any desired location in three-dimensional space such that the axis A<b>3</b> of the x-ray source <b>104</b> is substantially perpendicular to the surface of the x-ray detector <b>102.</b>

0227Returning to the discussion of step S<b>4402,</b> if the thicknesses of the lingual plate <b>4204</b> and the buccal plate <b>4206</b> in the mandible bone are desired, the x-ray source <b>104</b> should preferably be positioned below the patient's jaw, as illustrated in <figref idref="f0056 f0057 f0058">FIGS. 46-48</figref>. With this configuration, the teeth and the mandible (including the lingual plate <b>4204</b> and the buccal plate <b>4206)</b> comprise the object <b>50</b> and sub-objects illustrated in <figref idref="f0001">FIG. 1A</figref>. In an exemplary embodiment, the x-ray source <b>104</b> is initially positioned at the 0° position in the scan angle, which typically corresponds to a middle position in the scanning range (S<b>4404</b>). However, the x-ray source <b>104</b> may be initially positioned at any location within the scanning range. As shown in <figref idref="f0057">FIG. 47</figref>, it is preferable that a plane corresponding to the surface of the x-ray detector <b>102</b> is orthogonal to the imaging direction of the x-ray source <b>104.</b> An aiming device <b>4800</b> may be provided to aid with the alignment of the x-ray source <b>104</b> relative to the x-ray detector <b>102,</b> as illustrated in <figref idref="f0058">FIG. 48</figref>.

0228<figref idref="f0058">FIG. 48</figref> shows the x-ray source <b>104</b> which has been positioned with respect to the x-ray detector <b>102</b> with the assistance of an aiming device <b>4800.</b> The aiming device <b>4800</b> includes an arm <b>4804</b> which is connected to the x-ray detector <b>102</b> and the aiming ring <b>4806.</b> The connections between (i) the arm <b>4804</b> and the x-ray detector <b>102</b> and (ii) the arm <b>4804</b> and the aiming ring <b>4806</b> may be made by a mechanical connection (e.g., a snap-fit connection or a friction fit connection), a magnetic connection, an electrical connection, a chemical connection (e.g., glue), or a combination thereof. For example, the cross-sectional dimension of the arm <b>4804</b> may be substantially the same as an opening (see opening <b>4904</b> in <figref idref="f0059">FIG. 49</figref>) in the aiming ring <b>4806</b> that receives the arm <b>4804,</b> such that there is a significant amount of friction between the opening and the arm <b>4804</b> when the arm <b>4804</b> is inserted into the opening, thus establishing a friction fit. The amount of friction is sufficient to ensure that the aiming ring <b>4806</b> does not move during the radiographic scan. The arm <b>4804</b> may also extend through an opening so that the distance between the aiming ring <b>4806</b> and the x-ray detector <b>102</b> is adjustable and may be appropriately set by an operator. Once that distance is set, the friction between the opening and the arm <b>4804</b> will prevent the aiming ring from moving further. An opposite end of the arm <b>4804</b> may include a holder that is configured to receive the x-ray detector <b>102</b> via a snap-fit connection.

0229The aiming ring <b>4806</b> may also include a plurality of alignment markers <b>4902,</b> as shown in <figref idref="f0059">FIG. 49</figref>. The alignment markers <b>4902</b> may be formed from a known material whose attenuation factor is known. The alignment markers <b>4902</b> may be visible in the recorded two-dimensional x-ray projection image, and can serve as reference markers during the image processing. The alignment markers <b>4902</b> may also be used to measure the system geometry (i.e., the spatial relationship between the x-ray source and the x-ray detector). While a particular geometry may be assumed, the actual geometry may differ from the assumed geometry, due to misalignment of the x-ray source, vibration or flexure of the arm of the aiming ring (or the structure on which the x-ray source is mounted), or motion of the patient. Since, in one embodiment, the system geometry may determine which backprojection matrix is used and may also modify the filtering step in a filter backprojection, the positions of the alignment markers <b>4902</b> in the recorded image can provide useful information regarding the system geometry. As noted above, the aiming ring <b>4806</b> may also be provided with one or more openings <b>4904</b> through which the arm <b>4804</b> may be inserted to connect the aiming ring <b>4806</b> to the x-ray detector <b>102.</b>

0230A similar process may be performed if the thicknesses of the lingual plate <b>4208</b> and the buccal plate <b>4210</b> in the maxilla are desired. In this case, the x-ray source <b>104</b> may preferably be positioned above the patient's head, as illustrated in <figref idref="f0060 f0061 f0062">FIGS. 50-52</figref>. Like with a scan of lingual plate <b>4204</b> and the buccal plate <b>4206</b> in the mandible, the x-ray source <b>104</b> may preferably be positioned such that the imaging direction defined by the axis of the x-ray source <b>104</b> is substantially orthogonal to the plane of the x-ray detector <b>102,</b> as illustrated in <figref idref="f0061">FIG. 51</figref>.

0231As shown in <figref idref="f0062">FIG. 52</figref>, an aiming device <b>5200</b> may also be used to aid in the alignment of the x-ray source <b>104</b> with the x-ray detector <b>102.</b> Like aiming device <b>4800,</b> aiming device <b>5200</b> includes an arm <b>5204</b> which is connected to the x-ray detector <b>102</b> and the aiming ring <b>5206.</b> One or more of the components of the aiming device <b>5200</b> may be the same as the components of the aiming device <b>4800.</b> The components of the aiming device <b>5200,</b> however, may also be different from those included in aiming device <b>4800,</b> and specifically constructed for use in imaging the lingual plate <b>4208</b> and the buccal plate <b>4210</b> of the maxilla from an occlusal direction. Like with aiming device <b>4800,</b> the connections between (i) the arm <b>5204</b> and the x-ray detector <b>102</b> and (ii) the arm <b>5204</b> and the aiming ring <b>5206</b> may be made by a mechanical connection (e.g., a snap-fit connection or a friction fit connection), a magnetic connection, an electrical connection, a chemical connection (e.g., glue), or a combination thereof. Like with aiming device <b>4800,</b> the distance between the aiming ring <b>5206</b> and the x-ray detector <b>102</b> may be adjustable.

0232With the x-ray source <b>104</b> properly aligned with the x-ray detector <b>102,</b> the computer system <b>106</b> initiates the scanning operation to record a plurality of projection images in step S<b>4406</b>. As discussed above, on-board motor controller <b>120</b> controls the motorized stage <b>118</b> so as to translate the x-ray source <b>104</b> through a plurality of locations within the scan angle <b>112,</b> as illustrated in <figref idref="f0056">FIG. 46</figref>.

0233<figref idref="f0056">FIG. 46</figref> illustrates the translation (and rotation) of the x-ray source <b>104</b> during an imaging operation over a scan angle of X degrees (which may, for example, be ±20°). In <figref idref="f0056">FIG. 46</figref>, the x-ray source <b>104</b> translates along the motorized stage <b>118,</b> but also rotates such that the x-rays <b>110</b> emitted from each of the different locations converge substantially at a focal spot <b>122,</b> as illustrated in <figref idref="f0001">FIG. 1A</figref>. Of course, in another embodiment, the x-ray source <b>104</b> may be moved in a curved scan path <b>132</b> or a circular scan path <b>134,</b> as described above.

0234At each location, the x-ray detector <b>102</b> records the intensity of the x-rays <b>110</b> received from the x-ray source <b>104</b> and converts those x-rays <b>110</b> into electrical signals. The computer system <b>106</b> then processes those electrical signals to generate a two-dimensional projection image. The x-ray source <b>104</b> then moves to the next position and a subsequent two-dimensional projection image is recorded. This process repeats at each desired scan location over the scanning range to generate a plurality of two-dimensional projection images.

0235In step S<b>4408,</b> the computer system <b>106</b> processes the plurality of projection images to reconstruct a series of two-dimensional tomosynthesis image slices that extend in a depth direction (the Z-axis in <figref idref="f0056 f0057 f0058 f0059 f0060 f0061 f0062 f0063 f0064 f0065 f0066 f0067 f0068 f0069 f0070">FIGS. 46-38</figref> and <figref idref="f0060 f0061 f0062">50-52</figref>), in the manner described above.

0236In step S<b>4410,</b> a particular slice corresponding to a desired depth in the depth direction may be selected. As discussed above, an operator may wish to know the thickness of the lingual and buccal plates at a particular depth or depths. For example, as illustrated in <figref idref="f0063">FIGS. 53A and 53B</figref>, the operator may wish to know the thicknesses at depths D₁ and D₂, which are within the depth of the tooth cavity. The operator may input one or more desired depths via the input unit <b>114</b> (or the display unit <b>108</b> if a touchscreen), and the computer system <b>106</b> may return a two-dimensional image corresponding to the desired depth, as illustrated in <figref idref="f0063">FIG. 53C</figref> for a depth D₁. Unlike a traditional two-dimensional x-ray image, the two-dimensional image shown in <figref idref="f0063">FIG. 53C</figref> is substantially parallel to the plane of the x-ray detector <b>102</b> at the depth D₁, and thus generally perpendicular to the lingual and buccal plates. As such, the thicknesses of the lingual and buccal plates can be determined by direct measurement using the two-dimensional image (S<b>4412</b>).

0237More particularly, in step S<b>4412</b>, an operator may select two points in the two-dimensional image and the computer system <b>106</b> will automatically return the distance between those points in a desired unit. In one embodiment, the operator may draw a line between two points (as illustrated in <figref idref="f0063">FIG. 53C</figref>) and the computer system <b>106</b> will return the distance between the end points of the line. For example, assume an operator enters a depth D₁, corresponding to a depth which is roughly one-third of the depth of the tooth cavity shown in FIGS. 52A and 52B (Note: the trabecular region is omitted from FIGS. 52A and 52B for simplicity purposes). The computer system <b>106</b> will return and display the two-dimensional image shown in <figref idref="f0063">FIG. 53C</figref> on a display unit <b>108.</b> In <figref idref="f0063">FIG. 53C</figref>, the lingual plate <b>4204</b> and the buccal plate <b>4206</b> are clearly visible. As shown in <figref idref="f0063">FIG. 53C</figref>, an operator may draw lines <b>5302</b> and <b>5304</b> across the lingual plate <b>4204</b> and the buccal plate <b>4206,</b> respectively, and the computer system <b>106</b> will calculate and display the length of those lines (<b>5302</b> and <b>5304</b>) in order to determine the thicknesses (T_L and T_B) of the lingual plate <b>4204</b> (T_L) and the buccal plate <b>4206</b> (T_B) at that depth. By redrawing lines <b>5302</b> and <b>5304</b> at another location, the operator may determine the thicknesses of the lingual plate <b>4204</b> and the buccal plate <b>4206</b> at that location as well. If the operator wishes to make another thickness measurement at a second depth (e.g., depth D<b>2</b> shown in <figref idref="f0063">FIGS. 53A and 53B</figref>), the operator need only provide the tomosynthesis system <b>100</b> with the desired depth information via the input unit <b>114,</b> and the computer system <b>106</b> will return a corresponding two-dimensional image at that depth from which the operator can select points or draw a line so as to measure the thicknesses of the lingual plate <b>4204</b> (T_L') and the buccal plate <b>4206</b> (T_B'). Thus, contrary to a traditional x-ray imaging, the above method allows the operator to determine the thicknesses of the lingual and buccal plates at a desired depth, without subjecting the patient to a relatively large dose of radiation, which they may receive from a CBCT scan.

0238As noted above, the tomosynthesis system <b>100</b> may receive guidance from the operator indicating a clinical aspect of interest, which may be the lingual and buccal plates. To aid the operator in identifying the lingual and buccal plates at a particular depth, the computer system <b>106</b> may analyze the tomosynthesis stack and automatically segment the anatomical features therein. In other words, the computer system <b>106</b> may automatically determine which portion of one or more two-dimensional images correspond to the clinical aspect of interest (i.e., the lingual and buccal plates). In one embodiment, the computer system <b>106</b> may use relative attenuation factors, determined from analyzing the plurality of projection images, to discern the presence of one or more objects within the tomosynthesis stack and segment those objects from each other. The computer system <b>106</b> may then highlight, outline, or color code those segmented objects and display them on the display unit <b>108.</b> In one embodiment, the computer system <b>106,</b> having segmented the lingual and buccal plates from the surrounding objects, may automatically measure the thicknesses of those plates at one or more locations on the two-dimensional image and display those measurements on the display unit <b>108.</b> In an exemplary embodiment, a user may provide the computer system <b>106</b> with information regarding the location of a desired implant and a depth range for the implant. In return the computer system <b>106</b> may provide measurements of the plate thicknesses at one or more depths within that range. Of course, this does not preclude the operator from selecting a different set of points, or drawing a different line, at a desired location to measure the thickness of the lingual or buccal plate at that location.

Tracking Motion from Projection Images

0239The intraoral tomosynthesis system <b>100</b> will now be further described in conjunction with <figref idref="f0064">FIG. 54</figref>, which is a flow diagram illustrating a process of tracking motion of one or more objects using a plurality of projection images in an intraoral dataset according to an example embodiment herein. In one example embodiment, the process steps shown in <figref idref="f0064">FIG. 54</figref> can be embodied in a control program stored in a non-transitory computer-readable storage medium and loaded into the main memory <b>232</b> and/or the secondary memory <b>234</b> of the computer system <b>106</b> using the removable-storage drive <b>238,</b> the hard disk drive <b>236,</b> and/or the communications interface <b>246.</b> Control logic (software), when executed by the processor <b>222,</b> causes the computer system <b>106,</b> and more generally the intraoral tomosynthesis system <b>100,</b> to perform the procedures described herein.

0240Prior to starting the process illustrated in <figref idref="f0064">FIG. 54</figref>, the x-ray detector <b>102</b> and x-ray source <b>104</b> are first aligned. Typically, the x-ray detector <b>102</b> and the x-ray source <b>104</b> are manually aligned by a clinician. In that regard, an aiming device may be used to ensure a proper initial alignment. However, relative motion of the x-ray detector <b>102</b> and the x-ray source <b>104</b> may occur during a scanning operation as a result of one or more factors including, for example, movement of object <b>50</b> (e.g., a patient), sub-object <b>52</b> (e.g., an object inside the patient), x-ray source <b>104,</b> and/or x-ray detector <b>102.</b><figref idref="f0064">FIG. 54</figref> describes an exemplary method for tracking such relative motion.

0241In Step S<b>5402</b>, the intraoral tomosynthesis system <b>100</b> acquires a plurality of projection images of object <b>50</b> over a scan angle range <b>112</b> (which may be predetermined), including the orthogonal projection image, in the manner described above. For example, the x-ray source <b>104</b> is moved by the motorized stage <b>118</b> and control circuitry <b>120</b> to different positions within the scan angle <b>112,</b> and the computer system <b>106</b> controls the x-ray source <b>104</b> to emit x-rays <b>110</b> at each position. As discussed above, and shown in <figref idref="f0001">FIG. 1A</figref>, the x-ray source <b>104</b> may translate as well as rotate about a pivot point. In one example embodiment herein, the scan angle <b>112</b> ranges from -20° to +20°, with the scanning positions evenly distributed in increments of 0.8° to provide <b>51</b> scan angles, including the 0° position, although this example is not limiting. The x-rays <b>110</b> then pass through and are attenuated by the object <b>50</b> before being projected onto the x-ray detector <b>102.</b> The x-ray detector <b>102</b> converts the x-rays <b>110</b> into electrical signals (either directly or indirectly, as described above) and provides the electrical signals to the computer system <b>106.</b> The computer system <b>106</b> processes the electrical signals collected at each scan angle position to acquire the plurality of projection images, each image comprising an array of pixels. The image acquired with the x-ray source <b>104</b> at the 0° position is also referred to herein as an orthogonal projection image.

0242In one example embodiment herein, the color depth of each pixel value of the projection images may be 12-bit grayscale, and the dimensions of the projection images correspond to the standard dental size of the x-ray detector <b>102,</b> as described above. For example, a Size-2 detector may produce projection images that are approximately 1700×2400 pixels in size, a Size-1 detector may produce projection images that are approximately 1300×2000 pixels in size, and a Size-0 detector may produce projection images that are approximately 1200× 1600 pixels in size.

0243In Step S<b>5402,</b> the computer system <b>106</b> generates a sub-sample of projection images from the plurality of projection images obtained in step S<b>5402</b>. The sub-sample of projection images may be stored in, for example, main memory <b>232</b> or secondary memory <b>234.</b> The sub-sample of projection images is for use in a coarse image reconstruction, performed in step S<b>5406</b>. By using a sub-sample of projection images, less computational resources and time are consumed by the image reconstruction process in step S<b>5406</b>. Of course, in an alternate embodiment, step S<b>5404</b> may be skipped and the reconstruction process in step S<b>5406</b> may be conducted using all of the projection images obtained in step S<b>5404</b>.

0244The sub-sample of projection images may include less than all of the projection images obtained in step S<b>5402</b> (e.g., 25 out of 51 projection images may be included in the sub-sample of projection images). In one embodiment, for example, every other projection image in the plurality of projection images is included in the sub-sample of projection images. Preferably, at least half of the projection images are included in the sub-sample of projection images. Each individual projection image in the sub-sample of the projection images may be further sampled to reduce the number of pixels therein. For example, in one embodiment, half of the pixels in each projection image are filtered out (e.g., every other pixel in the projection image). This further reduces the computational intensity of the reconstruction process in step S<b>5406</b>. Of course, in another embodiment, all of the plurality of projection images may be retained for use in the reconstruction process in step <b>S5406,</b> but the number of pixels in each projection image may be reduced.

0245In step S<b>5406,</b> image reconstruction is performed using the sub-sample of projection images obtained in step S<b>5404</b>. This image reconstruction is considered a coarse reconstruction because less than all of the projection images and/or pixels are used in the reconstruction process. Computer system <b>106</b> may also perform deblurring and other image enhancements, as will be described further herein.

0246As discussed above, each reconstructed tomosynthesis image slice is comprised of an array of pixels that represent a cross-section of object <b>50</b> in a plane that is parallel to the surface of the x-ray detector <b>102</b> (the x-y plane in <figref idref="f0001">FIG. 1A</figref>). The plane is located a certain distance from the x-ray detector <b>102</b> in a direction that is orthogonal to the surface of the x-ray detector <b>102</b> (the z-axis direction in <figref idref="f0001">FIG. 1A</figref>).

0247Each tomosynthesis image slice has a certain thickness along the z-axis that is a function of the reconstruction technique used to create the tomosynthesis image slices and aspects of the geometry of the system <b>100,</b> including, primarily, the scan angle <b>112.</b> For example, each tomosynthesis image slice may have a slice thickness of 0.5 mm by virtue of the geometry of the system <b>100</b> and the reconstruction technique. The desired location of each tomosynthesis image slice along the z-axis is provided as an input to the reconstruction performed in Step S<b>5406</b> either as a pre-programmed parameter in computer system <b>106</b> or by user input via input unit <b>114</b> and/or display unit <b>108.</b> By example only, the computer system <b>106</b> can be instructed to reconstruct, from the plurality of projection images, a first tomosynthesis image slice that is one millimeter (1 mm) away from the surface of x-ray detector <b>102</b> along the z-axis, a last tomosynthesis image slice that is fifteen millimeters (15 mm) away from the surface of the x-ray detector <b>102,</b> and one or more image slices between the first image slice and the last image slice at regular increments along the z-axis of two-hundred micrometers (200 µm), for a total of seventy-one image slices.

0248Reconstruction of the tomosynthesis image slices in Step S<b>5406</b> may be performed in accordance with any existing or later developed reconstruction techniques. One exemplary technique which may be used is the shift-and-add technique. The shift-and-add technique utilizes information about the depth of sub-object(s) <b>52</b> along the z-axis that is reflected in the parallax captured by the plurality of projection images, as described above. According to this example embodiment, an image slice is reconstructed by first spatially shifting each projection image by an amount that is geometrically related to the distance between the image slice and the tomographic focal spot <b>122</b> along the z-axis. The shifted projection images are then averaged together to result in the image slice, where all sub-objects <b>52</b> in the plane of the image slice are in focus and sub-objects <b>52</b> outside of that plane are out of focus and blurry. This shift-and-add process is repeated for each image slice to be reconstructed. In the case of the image slice corresponding to the x-y plane that includes the focal spot <b>122,</b> the projection images are averaged together without first shifting because sub-objects <b>52</b> are already in focus for that plane.

0249As mentioned above, a deblurring technique may be used to deblur reconstructed tomosynthesis image slices. In one example embodiment herein, a deblurring technique that substantially reduces or removes blurry, out-of-plane sub-objects from an image slice can be performed in conjunction with the reconstruction technique (whether shift-and-add or another technique). Examples of deblurring techniques that can be employed are, for example, spatial frequency filtering, ectomography, filtered backprojection, selective plane removal, iterative restoration, and matrix inversion tomosynthesis, each of which may be used in Step S<b>5404</b> to deblur images reconstructed by the shift-and-add reconstruction technique (or another reconstruction technique, if employed).

0250In another example embodiment herein, Step S<b>5406</b> also can include the computer system <b>106</b> performing further automated image enhancements such as, for example, image sharpening, brightness optimization, and/or contrast optimization, on each reconstructed (and deblurred, where deblurring is performed) image slice in a known manner.

0251Additionally, in another example embodiment herein, the dimensions, position, and orientation of each image slice reconstructed in Step S<b>5406</b> are the same as the corresponding characteristics of the orthogonal projection image. Thus, when tomosynthesis image slices (or portions thereof) and the orthogonal projection image are overlaid over one another, corresponding anatomical features appearing in the images will be overlapped and aligned without scaling, rotation, or other transformation of the images.

0252In Step S<b>5408</b>, the computer system <b>106</b> processes the plurality of projection images acquired in step S<b>5402</b> to identify one or more objects in each projection image that may be appropriate for tracking motion. This process is detailed in <figref idref="f0065">FIG. 55</figref>.

0253As noted above, each projection image is comprised of a plurality of pixels, each with an assigned color depth (e.g., 12-bit grayscale). The total number of pixels in each projection image is dependent upon the size of the x-ray detector <b>102,</b> as discussed above. In a preferred embodiment, each of the projection images acquired in step S<b>5402</b> has the same number of pixels contained therein. In step S<b>5502</b>, a difference between neighboring projection images is calculated on a pixel-per-pixel basis, as shown in <figref idref="f0066 f0067">FIGS. 56A-57B</figref>.

0254<figref idref="f0066">FIG. 56A</figref> shows a first projection image <b>5600</b> recorded at a first scanning position in the scan angle <b>112.</b><figref idref="f0066">FIG. 56B</figref> is a second projection image <b>5602</b> recorded at a second scanning position in the scan angle <b>112.</b> The first and second scanning positions are adjacent to one another in the scan angle <b>112.</b><figref idref="f0067">FIG. 57A</figref> is a difference image <b>5700,</b> representing the difference between pixel values at the same x and y locations in the first and second projection images. The pixel values in the difference image (representing the difference between the first and second projection images) may be label as a set of values D₂(x,y). While <figref idref="f0067">FIG. 57A</figref> shows a difference image <b>5700</b> corresponding to the first and second projection images, computer system <b>106</b> calculates a difference image for each adjacent pair of projection images (that is projection images corresponding to adjacent scanning positions) to create a series difference images with pixel values D_ij(x,y).

0255The difference images calculated in S<b>5502</b> may be used to identify objects for motion tracking. Alternative methods may be used to identify the objects for motion tracking. For example, a thresholding image method (S<b>5504-S5506</b>) or a gradient segmentation method (S<b>5508-S5514</b>) may be used. Briefly, in the thresholding image method (S<b>5504-S5506</b>), described in further detail below, a difference image <b>5700</b> is used to generate a binary image <b>5700</b> (see <figref idref="f0067">FIG. 57B</figref>), which is subsequently analyzed to determine connected segments. One or more of the connected segments (e.g., the four largest connected segments) are identified as objects for motion tracking. The computer system <b>106</b> then generates a region of interest (ROI) around the identified objects (S<b>5516</b>).

0256In the gradient segmentation method (S<b>5508-S5514</b>), described in detail below, a gradient segmentation operation is performed on the difference image <b>5700</b> (S<b>5508</b>). The gradient segmentation operation calculates the magnitude and direction of the gradient at each pixel location in the difference image <b>5700.</b> The computer system <b>106</b> isolates the vertical and horizontal edges (S<b>5510</b>), based on the calculated gradient, and identifies the vertical and horizontal edges with a gradient magnitude above a certain magnitude (S<b>5512</b>). Of those vertical and horizontal edges, certain edges are selected as objects for motion tracking (S<b>5514</b>). The computer system <b>106</b> then generates a region of interest (ROI) around the identified objects (S<b>5516</b>).

0257Turning to the thresholding image method, in step S<b>5504</b> a thresholding operation is performed on the pixel values in the difference image D_ij(x,y) to produce a binary image. More specifically, each pixel value in D_ij(x,y) is compared to a threshold value. If the pixel value is greater than threshold value, then a pixel value at a corresponding point in a binary image (represented by a series of values B_ij(x,y)) is set to 1, otherwise the pixel value at the corresponding point is set to 0. In one embodiment, the threshold value may be set as the average pixel value in the difference image, resulting in half of the pixels in the binary image having a value of 1, and the other half having a value of zero. An exemplary binary image <b>5702</b> is shown in <figref idref="f0067">FIG. 57B</figref>.

0258Once the binary image <b>5702</b> is generated in step S<b>5504</b>, the binary image <b>5702,</b> or more specifically the values B_ij(x,y) comprising the binary 5702 image, are analyzed to determine connected segments (step S<b>5506</b>). This may be accomplished by analyzing the binary image <b>5702</b> to determine regions where values of "1" are continuously adjacent to one another. The largest connected segments in the binary image (i.e., the longest continuous chain of adjacent values of " 1") are then identified by computer system <b>106,</b> and one or more of those segments are identified as objects for motion tracking. In one embodiment, the computer system <b>106</b> identifies the four largest objects as objects for motion tracking; however, more or less objects may be used. As shown in <figref idref="f0067">FIG. 57B</figref>, the computer system <b>106</b> has identified four objects <b>5704-5707</b> for motion tracking from the binary image <b>5702.</b>

0259Turning to the gradient segmentation method, in step S<b>5508</b> the gradient of the difference image <b>5700</b> is calculated from the values D_ij(x,y). The computer system <b>106</b> analyzes the direction of the gradient on a pixel level basis to identify vertical and horizontal edges. For example, the computer system <b>106</b> analyzes adjacent pixels to determine if the direction of the gradient is substantially vertical or horizontal (e.g., within a range of 10 degrees of either vertical or horizontal). If a pixel falls outside of this range, it is disregarded. If the direction of the gradient is substantially vertical for adjacent pixels, then the adjacent pixels may correspond to a horizontal edge. If the direction of the gradient is substantially horizontal then the adjacent pixels may correspond to a vertical edge. The computer system <b>106</b> may determine the presence of a vertical or horizontal edge if the gradient for a minimum number of adjacent pixels is in the horizontal or vertical directions, respectively. In step S<b>5512,</b> the computer system <b>106</b> analyzes the identified vertical and horizontal edges to determine the magnitude of the gradient along those edges. In step S<b>5514,</b> the computer system <b>106</b> identifies vertical and horizontal edges where the magnitude of the gradient is above a certain threshold. For those vertical and horizontal edges where the magnitude of the gradient is above the threshold, the computer system <b>106</b> determines the size of the edges and selects one or more of the largest edges as objects for motion tracking. In one embodiment, edges comprised of 200 or more pixels are used for motion tracking.

0260Regardless of whether the objects for motion tracking are determined by the thresholding image method or the gradient segmentation method, once the objects for motion tracking are identified, the computer system <b>106</b> automatically generates regions of interest (ROIs) around the objects, in step S<b>5516.</b> For example, <figref idref="f0068">FIG. 58</figref> is a projection image where four regions of interest <b>5801-5804</b> have been generated around the positions of objects <b>5704-5707</b> in <figref idref="f0067">FIG. 57B</figref>.

0261After Step S<b>5408</b> is performed, control passes to step S<b>5410</b> in which the computer system <b>106</b> applies a tracking algorithm for each region of interest. The tracking algorithm is explained in further detail below, with reference to <figref idref="f0069">FIG. 59</figref>. In one embodiment, one or more processes (e.g., a filtering operation) may be performed prior to executing the tracking algorithm. For example, the pixel values in projection images may be filtered.

0262<figref idref="f0069">FIG. 59</figref> shows two projection images (<b>5902</b> and <b>5904</b>) recorded at adjacent scanning positions within the scan angle <b>112.</b> As shown in <figref idref="f0069">FIG. 59</figref>, the position of object <b>5906</b> is shifted up and to the left in the second projection image <b>5904.</b> To determine the amount of shift, a region of interest <b>5908</b> may be generated over a portion of object <b>5906</b> in the first projection image <b>5902.</b> In one embodiment, the region of interest <b>5908</b> may be windowed to reduce edge effects. The computer system <b>106</b> then overlays the region of interest <b>5908</b> at a first position (P₁) in the second projection image <b>5904,</b> and multiplies the pixels value from the region of interest <b>5908</b> in the first projection image <b>5902</b> (P_ROT(x,y)) by the pixels values corresponding to the area of the second projection image which the region of interest <b>5908</b> overlays (P_5902-P1(x,y)), respectively, and then sums the result. The sum is a cross-correlation function. The summation (i.e., the cross-correlation function) is recorded and stored in memory <b>232</b> or <b>234.</b> The region of interest <b>5908</b> is then translated to a second position and another multiplication is calculated and the corresponding summation is recorded. In one embodiment, the region of interest <b>5908</b> is translated by a unit of one pixel in either the x or y direction. However, the region of interest <b>5908</b> may be translated by a greater amount to reduce computational time. The region of interest is thus translated across the second projection image <b>5904,</b> and a series of summations (of the plurality of individual multiplications) are recorded. When the region of interest <b>5908</b> is overlaid with the same portion of the object <b>5906</b> that the region of interest <b>5908</b> corresponds to in the first projection image <b>5902</b> (at an alignment position P_A), the resulting multiplication produce a relatively large result, and thus the summation of those multiplications is also relatively large and identifiable by the computer system <b>106.</b> In other words, the computer system <b>106</b> determines the translation for which the cross-correlation function is a maximum as the alignment position. For example, assume the projection image is comprised of nine pixels arranged in a 3×3 array, with the center pixel having a value of pure white (e.g., <b>255</b>) and the remaining pixels have a value of pure black (i.e., 0). If the ROI is the center pixel, then the pixel value for the ROI is also pure white (e.g., 255). Note, since the ROI only contains one pixel, there is only one multiplication operation and no summation operation in this example. The result of a multiplication of the pixel value corresponding to the ROI (255) by any pixel other than the center pixel (0) is 0. When the pixel value corresponding to the ROI (255) is multiplied by the center pixel (255), however, the result is 65,025. This result is a maximum among the calculations and thus corresponds to an alignment position. Since the translations of the ROI are approximately known, the possible alignment position is restricted to a set of possible alignment positions. In another embodiment, the cross-correlation function may be determined by applying a fourier transform technique.

0263With the alignment position and the position of the region of interest in the first projection image known, the computer system <b>106</b> may determine how much shift occurred in both the x and y directions between the first projection image <b>5902</b> and the second projection image <b>5904.</b>

0264As noted above, the computer system <b>106</b> may track the motion of more than one object. As such, the above tracking algorithm may be performed for each ROI corresponding to the objects for motion tracking (e.g., objects <b>5704-5707</b> in <figref idref="f0067">FIG. 57B</figref>). If the motion of more than one object is tracked, then (in step S<b>5412</b>) an average pixel shift may be determined from the calculated shifts for each ROI, with the average pixel shift along the x and y axes calculated separately.

0265Through the above-described process, an average shift for each ROI between two projections images may be determined. This process may be used to calculate the average pixel shift for ROIs between projection images obtained at adjacent scanning positions across the scan angle <b>112.</b> For example, an average ROI shift (ROI₁₂) between a first projection image (corresponding to scan angle of 20°) and a second projection image (corresponding to scan angle of 19.2°) may be calculated, as well as an average ROI shift (ROI₂₃) between the second projection image (corresponding to the scan angle of 19.2°) and a third projection image (corresponding to a scan angle of 18.6°).

0266Once all of the ROI shifts are calculated across the scan angle <b>112,</b> it is necessary to shift each ROI in each projection image by the cumulative shift amount in order for all of the ROIs to align in the tomosynthesis stack.

0267The cumulative shift is the sum of the ROI shifts that occur in the x and y direction between the orthogonal imaging position (corresponding to 0° in the scan angle, e.g. SP_0x) and a particular scanning position. For example, assume there are 51 scanning positions within the scan angle <b>112.</b> The cumulative ROI (along the x-axis) shift for scanning position SP₊₂ (corresponding one end of the scan angle <b>112</b>), is the sum of the cumulative ROI shifts between scanning positions SP_+25x, SP_+24x, SP_+23x, . . . , to SP_+1x. In a similar manner, the cumulative ROI shift for scanning position SP₊₁₃, along the x axis is the sum of the cumulative shifts between scanning positions SP₊₁₃, SP_+12x, . . . , to SP_+1x. In step S<b>5414,</b> the computer system <b>106</b> shifts each ROI in each projection image by the appropriate cumulative shift amount.

0268After Step S<b>5414,</b> control passes to Step S<b>5416</b> which takes as input the coarse tomographic image slices reconstructed in step S<b>5404</b> and the shifted ROIs of step S<b>5414.</b> In step S<b>5416,</b> the computer system <b>106</b> determines the plane of focus for each object used for motion tracking using, for example, the focus technique described above. In one embodiment, computer system <b>106</b> evaluates how in focus the ROI is for each slice in the tomosynthesis stack and assigns a corresponding focus factor. The computer system <b>106</b> determines the tomosynthesis slice with at least a local maximum focus factor which corresponds to the plane in which the object is in focus. By determining the plane in which the object is in focus, the location of the object relative to the surface of x-ray detector <b>102</b> and the x-ray source <b>104</b> can be determined. In other words, the distance in the imaging direction (the z-axis) from tomosynthesis image slice that contains a high-focus image of the object to the x-ray source <b>104</b> is used as the distance in the imaging direction from the object(s) for motion tracking to the x-ray source <b>104.</b> With this information, the actual shift amount for the x-ray source between two projection images corresponding to adjacent scanning positions may be determined (step S<b>5418</b>).

0269It should be noted that the same object may not be used to determine the actual shift amount for each pair of projection images. An object used for motion tracking between a first and second projection image, may not be the same object used for motion tracking between a second and third projection image. If a different object is used, then the distance from the x-ray source <b>104</b> to the different object in the imaging direction will likely change. Nevertheless, the above process takes this fact into consideration, and thus produces actual shift amounts for the x-ray source between the projection images regardless of whether the same object is used for motion tracking or not.

0270In step S<b>5420,</b> the actual shift amounts for the x-ray source may be compared to theoretical shift amounts. As mentioned above, motion of the patient, the x-ray detector <b>102,</b> and/or the x-ray source <b>104,</b> may cause the actual shift amounts to depart from the theoretical shift amount, which assumes that the system geometry (the spatial relationship between the patient, x-ray detector <b>102,</b> and x-ray source <b>104</b>) is constant. However, this comparison is not necessary to generate motion-compensated tomographic images, and is therefore an optional step.

0271In step S<b>5422,</b> the scan angles (which are one input for the reconstruction algorithm) are updated based on the actual shift amounts for the x-ray source <b>104</b> for motion tracking. Corrections to the scan angles may be calculated based on the actual shift amounts and the pixel size (in mm) for a given x-ray detector <b>102.</b> The updated scan angles are then input into the reconstruction algorithm (step S<b>5424</b>). The computer system <b>106</b> then performs image reconstruction on the motion-compensated projection images (step S<b>5426</b>).

0272The motion-compensated tomographic images may then be displayed on display unit <b>108.</b> In one example embodiment herein, the displaying can be performed as to show the entire stack, or one or more selected image slices of the stack, using display unit <b>108,</b> and interactive controls (e.g. via display unit <b>108</b> and/or input device <b>114</b>) enable a user to select between those two options, and to select one or more image slices for display, and also to select one or more particular regions of interest in the image(s) for display (whether in zoom or non-zoom, or reduced fashion). In a further example embodiment, stack controls are provided and can include a scroll bar, which enables the user to manually select which image slice is displayed on the display unit <b>108,</b> and/or can include selectable control items, such as play, pause, skip forward, and skip backward, (not shown) to enable the user to control automatic display of the tomosynthesis stack, as a cine loop for example, on the display unit <b>108.</b>

0273As will be appreciated by those of skill in the relevant art(s) in view of this description, the example aspects described herein can be implemented using a single computer or using a computer system that includes multiple computers each programmed with control logic to perform various of the above-described functions.

0274The various embodiments described above have been presented by way of example and not limitation. It will be apparent to persons skilled in the relevant art(s) that various changes in form and detail can be made therein (e.g., different hardware, communications protocols, and the like) without departing from the scope of the present invention. Thus, the present invention should not be limited by any of the above-described exemplary embodiments, but should be defined only in accordance with the following claims.

0275In addition, it should be understood that the attached drawings, which highlight functionality described herein, are presented as illustrative examples. The architecture of the present invention is sufficiently flexible and configurable, such that it can be utilized and navigated in ways other than that shown in the drawings.

0276Moreover, the example embodiments described herein are not limited to intraoral tomosynthesis imaging. The example embodiments described herein can be used to perform scans of other anatomical regions. In addition, one or more of the above techniques may also be applied to tomography in general (including CT and CBCT).