Artifact removal from nuclear image

The contents disclosed in this document generally relate to nuclear imaging, and more specifically, nuclear imaging (nuclear) such as single photon emission computed tomography (SPECT) or other emission tomography. It relates to the correction of artifacts caused by out-of-field sources in imaging technology.

Various imaging techniques are known and are currently used for medical diagnostic purposes and the like. Certain techniques, such as SPECT, within such techniques are usually administered in the form of radiopharmaceuticals that can be carried by the particular tissue of interest and, in some cases, bound to the particular tissue. It utilizes the emission of γ (gamma) rays generated during radioactive decay of isotopes (or radionuclides). In such a nuclear imaging technique, the emission is detected by an appropriate gamma ray detector. More specifically, a suitable gamma-ray detector can consist of multiple components that respond to incident radiation and generate image data proportional to the amount of radiation that hits individual regions within the detector. The image data generated by these detector components can then be reconstructed to generate an image of the internal structure of the subject.

Such a system has proven to be very useful in providing high quality images with sufficient diagnostic value, but further improvements are possible. For example, in some cases, a particular part of the patient's anatomy may be of interest to the clinician. In such cases, clinical physicians may seek to obtain imaging data for the organ of interest. However, due to the way data is collected (typically from multiple views or angles around the patient), some of the data collected is only in areas of interest to the clinician. Instead, it may contain data representing other intrapatient anatomical structures that may not be of interest. For example, other organs or regions of the patient may be involved in the disintegration of radiopharmaceuticals and therefore may emit gamma rays above the observed background levels in other ways. As long as these other organs or regions appear in the image data collected in a particular view, they can affect the quality of the images generated for the actual region of interest. This effect is present in cameras with a small field of view where the size of the imaging detector is limited, and may be especially present in cameras where the detector is aimed at the area of interest. Similarly, there may be problems with PET (Positron Emission Tomography) using non-perfect circular detectors.

The inventions disclosed herein are within the patient's anatomy within the acquired nuclear imaging data set (such as the SPECT data set or other emission-type imaging modality data sets). The present invention relates to a method for reducing or removing a portion related to a contribution from a portion other than the region of interest (that is, a radiation source outside the field of view). In certain embodiments, reconstruction methods such as the successive approximation reconstruction method can be used, in which unpredicted or undesired lines are used with modeled or predicted image data. Data contributions from the source can be reduced or eliminated. In this aspect, while generating an image corresponding to the data acquired from the area of interest of the patient (such as the heart), it relates to other areas that may be inadvertently imaged during the data acquisition process. The impact of the data can be reduced or eliminated.

According to one aspect of the invention disclosed herein, an image analysis system is provided. This image analysis system includes one or more processing components configured to receive a measured view that generally includes the organ or region of interest. At least some of these measured views include the contribution of radiation from outside the organ or region of interest. One or more processing components are also configured to execute one or more executable routines stored in memory. When executed, these stored routines reconstruct multiple measured views to produce the first image and use the first image to generate a corresponding estimated view for each measured view. Generate and compare each measured view with the estimated view to derive an indication of the presence of radiation contributions from outside the organ or region of interest in each measured view. The display is then used to reconstruct the final image with reduced or removed contributions of radiation from outside the organ or region of interest. The image analysis system also includes an interface circuit configured to allow the user's interactive interaction with the image analysis system.

According to another aspect, one or more machine-readable media in which multiple routines are encoded are provided. These routines, when executed by a processor, perform multiple actions, including the act of obtaining multiple different measured views. Multiple measured views depict the contribution of radiation from the organ or region of interest. One or more of the measured views also contains secondary radiation contributions from outside the organ or region of interest. The multiple actions performed also include the act of reconstructing the first image using multiple measured views, the act of partitioning the organ or region of interest in the first image to form a mask. , Also includes the act of reprojecting the first image with a mask to generate multiple cleaned views with the pixels outside the area defined by the mask set to the adjustment value. The final image is reconstructed at least partially based on multiple cleaned views.

According to yet another aspect, an image reconstruction method is provided. According to this method, multiple measured views representing the radiation emitted by an organ or region of general interest are obtained, as if viewed from different locations. One or more of the plurality of measured views includes radiation emitted by an out-of-sight source with respect to the field of view of at least some of the remaining plurality of measured views. A corresponding estimated view is generated for each measured view. Each measured view and corresponding estimated view is affected (ie, affected) by one or more within one or more measured views, including contributions from radiation emitted by out-of-field sources. Compared to identify areas. A final image is generated with the contribution of the identified affected area reduced or removed.

These and other features, aspects and advantages of the present invention will be better understood by reading the following detailed description with reference to the accompanying drawings. In the drawings, similar elements are represented by similar reference codes throughout the drawings.

<figref num="1">FIG. 1 is a schematic representation of an embodiment of a SPECT imaging system suitable for use in accordance with the present invention.</figref><figref num="2">FIG. 2 is a schematic diagram showing an example of SPECT image acquisition performed in various views using a plurality of pinhole camera type gamma detectors according to various aspects of the present invention.</figref><figref num="3">FIG. 3 is a schematic diagram showing an image acquired in the first view by the image acquisition configuration shown in FIG.</figref><figref num="4">FIG. 4 is a schematic diagram showing an image acquired in the second view by the image acquisition configuration shown in FIG.</figref><figref num="5">FIG. 5 is a schematic diagram showing an image acquired in the third view by the image acquisition configuration shown in FIG.</figref><figref num="6">FIG. 6 is a schematic diagram showing an image acquired in the fourth view by the image acquisition configuration shown in FIG.</figref><figref num="7">FIG. 7 is a schematic diagram showing an example of SPECT image acquisition performed in various views using a collimated gamma detector according to various aspects of the present invention.</figref><figref num="8">FIG. 8 is a schematic diagram showing an image acquired in the first view by the image acquisition configuration shown in FIG.</figref><figref num="9">FIG. 9 is a schematic diagram showing an image acquired in the second view by the image acquisition configuration shown in FIG.</figref><figref num="10">FIG. 10 is a schematic diagram showing an image acquired in the third view by the image acquisition configuration shown in FIG.</figref><figref num="11">FIG. 11 is a schematic diagram showing an image acquired in the fourth view by the image acquisition configuration shown in FIG.</figref><figref num="12">FIG. 12 is a first flow diagram of processor executable logic for dealing with image artifacts caused by out-of-field sources, according to various aspects of the invention.</figref><figref num="13">FIG. 13 is a schematic diagram illustrating an example of a view that depicts a suitable extension of the identified affected area within the view according to various aspects of the invention.</figref><figref num="14">FIG. 14 is a second flow diagram of processor executable logic for dealing with image artifacts caused by out-of-field sources, according to various aspects of the invention.</figref>

As described herein, the present invention relates to the generation of nuclear medicine images such as SPECT or other emission-type tomography reconstructions that reduce or eliminate the effects of out-of-field radiation sources. For example, in one embodiment a reconstruction method such as the successive approximation reconstruction method can be used, in which a predicted or modeled view of the actual organ or physiological part of interest is measured. It is compared with the actual measured view of the source-derived portion of the image data in the field of view. The portion of the image data affected by the out-of-field source can be rejected, replaced, corrected and downweighted as part of the reconstruction process. In this aspect, the image of the area of interest of the patient can be reconstructed to reduce or eliminate the effects of such out-of-field sources. This effect is present in cameras with a small field of view where the size of the imaging detector is limited, and may be especially present in cameras where the detector is aimed at the organ of interest. With a large field of SPECT camera where the entire cross section of the object to be imaged is within every individual view (sometimes called a "projection"), this is rarely a problem. However, the use of small field detectors can be essential or can improve cost effectiveness, and can also increase sensitivity, especially for detectors that are aimed at the organ of interest. , The resolution can be improved, the dose to the patient can be reduced, the throughput can be improved, and / or the patient's discomfort can be alleviated by reducing the acquisition time.

With the above description in mind, FIG. 1 shows a schematic diagram of an example of a SPECT imaging system suitable for using this technique. Not surprisingly, the various techniques described in this document may also be suitable for use in other emission-type tomography modality. System 10 in FIG. 1 is useful for subject 14 with a suitable detector component (such as a pinhole gamma camera with a semiconductor or scintillation detector or a collimated gamma camera) as detailed below. Designed to produce a variety of images. The subject is positioned within the scanner 16 and the patient support 18 is positioned within the scanner 16. The support can be mobile within the scanner so that different tissues or anatomical structures of interest 20 within the subject can be imaged. Prior to image data collection, a radioisotope, such as a radiopharmaceutical substance (also called a "radiotracer"), is administered to the patient, which can bind to or incorporate into a particular tissue or organ 20. Typical radioisotopes include elements in various radioactive forms, but many SPECT imaging are isotopes of technetium that emit gamma rays during decay.^99mIt is based on Tc). Various additional substances can be selectively combined with such radioisotopes to target specific areas or tissues 20 of the body.

Gamma rays emitted by radioisotopes are detected by a detector component 22 such as a digital detector or gamma camera. Although shown in the figure as a planar device placed above the patient for brevity, in practice the detector structure (s) 22 is placed around the patient, eg. Can be placed in an arc or ring around the patient, or the detector structure (s) 22 arc or orbit around the patient during data acquisition. It can be attached to a positioning device (eg, a C-shaped arm, gantry, or other movable arm) that allows it to move along or reorient to the patient during data acquisition. In general, the detector structure (s) 22 can typically detect gamma rays and otherwise generate a detectable signal in response to such radiation1 Includes one or more components or elements. In the illustrated embodiment, the detector structure has one or more collimators and one scintillator (collectively represented by reference numeral 24). The collimator can only emit gamma rays in a specific direction (typically perpendicular to the scintillator) so that they collide with the scintillator. Scintillators are typically made of crystalline materials such as sodium iodide (NaI), where scintillators receive γ-rays with relatively low energy (eg, in the ultraviolet region) photoenergy. Convert to. The photomultiplier tube 26 then receives this light and generates image data corresponding to photons that collide with specific individual pixel regions. In other embodiments, the detector structure 22 cannot collimate, but instead uses other gamma ray detection techniques, such as one or more pinhole gamma cameras as described herein. be able to.

In the illustrated embodiment, the detector structure (s) 22 is coupled to the system control and processing circuit 28. The circuit can include a number of physical and / or software components that can collaborate to collect and process image data to produce the desired image. For example, this circuit can include a raw data processing circuit 30 that can first receive data from the detector structure (s) 22 and then perform various filtering, value adjustments, and so on. The processing circuit 32 allows overall control of the imaging system and also allows the manipulation and / or reconstruction of the image. The processing circuit 32 can also perform a calibration function, a correction function, and the like on the data. The processing circuit 32 can also perform image reconstruction functions such as those based on known algorithms (eg, back projection, successive approximation reconstruction, etc.). Such functions can also be performed in post-processing on-site or in remote equipment. Not surprisingly, the various image reconstruction and artifact correction algorithms described herein can be partially or wholly embodied using one or both of the raw data processing circuits 30 and / or the processing circuits 32. it can.

In the illustrated embodiment, the processing circuit 32 interacts with a control circuit / interface 34 that allows control of the scanner and its components (including patient supports, cameras, etc.). Moreover, the processing circuit 32 may be used to store image data, calibration or correction values, routines performed by the processing circuit (such as the artifact correction algorithm described herein), etc. It is supported by various circuits such as. In one embodiment, the processing circuit implements one or more successive approximation reconstruction algorithms that can utilize techniques for reducing or eliminating the effects of out-of-field sources as described herein. Such sequential approximation reconstruction methods generally make repeated use of comparisons between predicted or modeled images and observed or measured image data to form the geometric structure of the imaging system. Artifacts or irregularities due to non-physiological factors such as factors associated with out-of-sight source effects, attenuation, scattering, etc. can be reduced. In such a successive approximation reconstruction method, the convergence process or loop can be repeated or repeated for a specified number of iterations or until some completion criterion such as cost function minimization is met.

Finally, the processing circuit can interact with an interface circuit 38 designed to assist the operator interface 40. The operator interface directs the imaging sequence, allowing the scanner and system settings to be viewed and adjusted, and the like. In an exemplary embodiment, the operator interface includes a monitor 42 capable of displaying the reconstructed image 12.

In a facility environment, the imaging system 10 allows the transfer of system data to and from the imaging system, as well as the transmission and storage of image data and processed images. Can be connected to the network. For example, local area networks, wide area networks, wireless networks, etc. can enable storage of image data in radiomedical department information systems and / or hospital information systems. Such network connections also allow the transmission of image data to remote post-processing systems, clinics and the like.

In one embodiment, a nuclear imaging system, such as the SPECT imaging system of FIG. 1, is configured to image a targeted or confined area of the patient's body, such as a particular organ of interest. Can be done. For example, some such systems can be configured for cardiac imaging and may include multiple pinhole or dedicated cardiac cameras with a small or limited field of view. In such an embodiment, the reconstituted volume is typically smaller than the patient's portion of the field of view of the camera and smaller than the organ or region of interest (eg, the heart or other organ). It can be made only slightly larger.

In a typical embodiment, image data can be obtained for a variety of different views with respect to the patient. As a result of these different views, one particular view was emitted outside the field of view associated with the other view and outside the region corresponding to the volume to be reconstructed (ie, the reconstructed volume). Gamma rays may be detected. When such image data from an out-of-field source is present, the image reconstruction process is mathematically consistent because the data from an out-of-field source cannot be reconstructed into the reconstruction volume. There is no. Such mathematical inconsistencies may manifest themselves as artifacts within the reconstructed image volume.

For example, in a cardiac camera embodiment, the liver may be found in the data corresponding to one or more angular views, but not in the other views. For example, reference to FIGS. 2-6 shows a configuration with a plurality of pinhole cameras 60, in which these pinhole cameras 60 have different viewing angles (labels) around the patient 14. Placed or moved between them (marked with "A"-"D"). Each pinhole camera 60 has an associated field of view 62 (each indicated by a broken line) from a given viewing angle, and that field of view attempts to acquire image data by the pinhole camera 60 at that angle. Corresponds to 14 parts of the patient. As will be appreciated, a pinhole camera 60 as shown generally acquires a conical projection image corresponding to an inverted image of the field of view 62 from each camera 60.

The illustrated example shows an example of cardiac imaging, where each field of view 62 includes the heart 64 of patient 14 (ie, the heart 64 constitutes an area or organ of interest). The combination of the plurality of fields of view 62 defines a reconstructed volume 66, which is the area or organ (generally of interest) for which image data is to be acquired by each view. In this example, it corresponds to the area including the heart). Note that, for the sake of brevity, Figure 2 shows four pinhole gamma cameras in one plane. In the geometric configuration of a real camera system, more similar cameras can be arranged in three dimensions, for example, not all views need to be in the same plane.

There may be one or more other organs or structures outside the reconstituted volume 66, which also emit gamma rays due to the excretion or circulatory function performed by those organs, etc. Can act as a place where you can. In the illustrated example, one such organ is the liver 70, which is visible within some of the fields of view 62 of each of the plurality of pinhole cameras 60, but not all of them. For example, with reference to FIGS. 3 to 6, various images 72,74,76,78 corresponding to the image data acquired by the respective pinhole type gamma cameras 60 are displayed in the respective views A to D. It is shown.

For example, FIG. 3 shows a stylized example of a cardiac image 72 acquired by the SPECT system configuration of FIG. In this example, image 72 in FIG. 3 is acquired by the pinhole camera 60 at view position A in FIG. The field of view 62 of each pinhole camera 60 from view position A covers the area or organ of interest (heart 64 in this example) and has no contribution from other sources.

Next, FIG. 4 shows a stylized example of another heart image 74 corresponding to the image acquired by the SPECT system configuration of FIG. 2 at view position B. As with each image 72 captured at view position A, the field of view 62 of each pinhole camera 60 at view position B is the area or organ of interest, eg, the heart 64 (from different perspective directions). Includes what you see) and has no contributions from other sources. However, image 72 captured in view A and image 74 captured in view B image the region of interest (eg, heart 64) from different viewing directions and therefore from different different viewing angles. It's different because it depicts the area of interest.

FIG. 5 shows a stylized example of another cardiac image 76 corresponding to the image acquired by the SPECT system configuration of FIG. 2 at view position C. Unlike images 72 and 74 acquired at view positions A and B, the field of view 62 of each pinhole camera 60 at view position C is not only the area or organ of interest (eg, heart 64), but also. It also includes gamma sources outside the region of interest (in this example, the liver 70, i.e., out-of-field gamma sources). In fact, in the illustrated example, the heart 64 and the liver 70 overlap in the perspective direction of the pinhole camera 60 at view position C. Thus, the image data 76 acquired by the pinhole camera 60 in view C contains contributions from out-of-field sources (eg, liver 70) and, as a result, is of interest to include the heart 64. Artifacts can occur when reconstructing volumes.

Similarly, FIG. 6 shows the last stylized example of cardiac image 78 corresponding to the image acquired by the SPECT system configuration of FIG. 2 at view position D. As in image 76, the field of view 62 of each pinhole camera 60 at view position D is of the region or organ of interest (eg, heart 64), as well as out-of-field sources (eg, liver 70). Includes both. However, unlike image 76, in cardiac image 78, heart 64 and liver 70 do not overlap in the field of view, but gamma-ray data associated with liver 70 is still of interest (eg, heart 64). May cause artifacts within the reconstructed volume 66 containing. However, even if the liver 70 and the heart 64 do not overlap in this example, radiation from the liver may adversely affect the reconstruction of the heart image in this view due to radiation scattering effects and the like. ..

The above example describes an example of an image acquisition configuration using the pinhole type camera 60, and the pinhole type camera 60 generally has an inverted image of the field of view 62 from the camera 60 as described above. Acquire the conical projection image corresponding to. In another embodiment, as shown in FIG. 7, a collimator-type detector assembly 90 or collimated camera can be used, which uses both a collimator and a panel-type detector in the assembly. A collimator within such an assembly 90 acts to limit the angular range of gamma rays incident on the detector panel, thereby helping to localize gamma ray emissions. In such an image acquisition configuration, the collimated detector assembly 90 has a limited non-inverted field of view 62 that does not expand with distance, unlike the pinhole camera configuration. As in the case of the pinhole camera image acquisition described with respect to FIGS. 2 to 6, the images acquired in various views A to D using the illustrated collimated detector assembly are outside the reconstruction volume 66. Image data from (eg, from liver 70) can be included.

FIG. 8 shows a stylized example of the cardiac image 96 obtained by the SPECT system configuration of FIG. In this example, image 96 is acquired at view position A in FIG. 7 using a collimated detector assembly 90. The field of view 62 of the collimated detector assembly 90 from view position A covers the area or organ of interest (heart 64 in this example) and has no contribution from other sources. Similarly, cardiac image 98 in FIG. 9 shows a representative stylized image as can be obtained by the collimated detector assembly 90 at view position B in FIG. 7, and therefore from different view angles. Indicates the area of interest (eg, heart 64).

Similarly, FIGS. 10 and 11 show images 100 and 102 that can be obtained using the collimated detector assembly 90 of FIG. 7 at view positions C and D, respectively. As shown in images 100 and 102, an out-of-field source of gamma rays (represented by liver 70 in this example) is present within the field of view 62 associated with a particular view position. Sometimes. As a result, out-of-field radiation sources are either separate from the region or organ of interest (eg, heart 64) or (see image 102 in FIG. 11) with respect to the image data acquired at those view positions. Contributions that overlap the area or organ of interest (see image 100 in Figure 10) may occur. As mentioned in previous examples, volumes reconstructed using image data containing contributions from such out-of-field sources may contain artifacts or other irregularities.

With this in mind, the data measured by the illustrated SPECT imaging system consists of a set of measured views that are generally formed by radiation emitted from the organ or region of interest. You will notice that. This set of measured views constitutes data that can be reconstructed to produce the volume of interest that contains the region or organ of interest.

As long as radiation from out-of-field sources can be present, such radiation typically contributes to (affects) a limited number of views within multiple measured views, and therefore. Not constant or uniform among multiple measured views. Radiation from such out-of-field sources, when present, may appear as a continuous or uniform area, and in some cases, radiation from out-of-field sources may be the organ, system of interest. A pattern may be formed depending on the geometrical composition of the. Moreover, radiation from such out-of-field sources may or may not overlap the organ or region of interest, depending on the measured view in question.

Radiation from out-of-field sources, when present, is greater than the observed background radiation and is comparable to or comparable to the count density associated with the organ or region of interest in the affected measured view. May produce higher count densities. As a result, it would be desirable to distinguish the measured count data associated with the out-of-field source from the measured count data associated with the organ or region of interest. This identification of data related to out-of-field sources can separate this data from the affected measured view and leave the signal associated with the organ or region of interest intact. it can.

With the above in mind, FIG. 12 will explain an example of an algorithm that can be embodied as an image processing control logic that can be executed in a system using a processor. In this example, N measured views 122 (V)_MEASURED) Is obtained by the SPECT system at various locations around the organ or region of interest of the patient (block 120). Each measured view 122 is a two-dimensional image, some or all of which include at least a portion of the organ or region of interest within the patient. In one example, some of the multiple measured views 122 also include image data generated in response to radiation emitted by one or more out-of-field sources.

Performing the first reconstruction for multiple measured views 122 (block 124), the first reconstruction image (R)_INITIAL) 126 is produced, eg, a reconstructed volume containing the region or organ of interest. In the embodiment of acquiring a plurality of measured views 122 using a conventional collimated camera, the first reconstruction can be performed using a filter-corrected backprojection or successive approximation reconstruction algorithm. In other embodiments where multiple pinhole cameras are used to obtain multiple measured views 122, the first reconstruction is performed using a successive approximation reconstruction algorithm or other suitable reconstruction algorithm. be able to. In the first reconstruction (block 124), the camera system matrix (CSM) 128 can be used in the reconstruction that relates the responses at various detector elements to the voxels in the field of view. In some embodiments, additional corrections or processing may be performed in connection with reconstructing the first image 126. For example, attenuation correction and / or scattering correction can be performed as part of or after the reconstruction of the first image 126. At its discretion, only a portion of the plurality of measured views 122 is utilized for the reconstruction of the first image 126. For example, views from positions known to have a relatively high probability of including contributions from out-of-field sources can be excluded.

The first image 126 is typically occupied by an organ or region of interest, but an artifact or irregularity that may result from the contribution of radiation from one or more out-of-field sources. May include. In one embodiment, the region or organ of interest is based on identifying contiguous areas (s) with high voxel values (eg, voxel values above a certain threshold) in the first image 126. Can be divided. In such an embodiment, the expansion process is also performed to slightly expand the area partitioned to include adjacent or neighboring pixels so as to include adjacent or neighboring pixels, which is of interest. It is possible to ensure that a certain area or organ is covered by the divided area. In some embodiments, a known internal organ atlas (map) can be used to identify the organs in the first reconstructed image. Such atlases are known in the art and are prepared for different organs of different types of patients and conditions. Organs within the atlas can be aligned to the first image (eg, by image movement, rotation, scaling and / or other deformation). The first image can then be replaced with the matched atlas organ. This method allows the defective organ in the first image (which may have lost physiologically non-functional parts) to be replaced with a matched full-sized organ.

If the organ or region of interest is identified and / or segmented in the first image 126, various optional actions can be taken based on this segmentation. For example, in one embodiment, voxels that have not been identified as being within the segment corresponding to the voxels in the region of interest (eg, voxels other than the organ of interest) will have the values of these voxels. By setting it to zero, it can be effectively removed from the first image 126. In another embodiment, the value of the voxel other than the organ of interest is set to a value corresponding to the average background value, such as the average voxel value for voxels that were not assigned to the organ or region segment of interest. be able to. In other embodiments, voxel values other than the organ of interest can be blunted and / or, if necessary, bound to positive voxel values. In other embodiments, none of these actions need to be performed.

In one embodiment, the first image 126 is reprojected based on the model or a priori prediction 134 (block 130) and N estimated views (V)._ESTIMATED) Generate 132. Typically, the model 134 used in the reprojection process 130 is defined based on known parameters of the imaging system and is performed based on or in the known structure of the imaging system (eg, known lines). It can be determined based on the results of experimental measurements (using a source or phantom). The model or other prediction 134 typically corresponds to the model or prediction used in the successive approximation reconstruction algorithm to systematically test and address deviations and / or artifacts in the reconstructed image. Can be done. In such reprojections, the camera system matrix 128 can be used for reprojection to adequately model the physical and geometric effects of the camera or detector assembly. Similarly, a suitable attenuation and / or scatter correction model can be used for the reprojection process as long as the attenuation and / or scatter correction has been previously used.

By comparing the measured view 122 with the estimated view 132, it is possible to identify the view or region within the view that may have been affected by out-of-field sources (block 140). In one embodiment, the affected (ie, affected) view or region is identified by identifying regions or pixels that differ between the measured view 122 and the estimated view 132 (ie, difference data 142). can do. By using statistical scales, the affected view or view portion can be determined. For example, pixels where the difference between the measured data and the projected data is greater than or equal to the threshold are considered affected. The threshold can be a preset fraction of the measured, other or projected value, or It can be related to the estimated noise (such as the standard deviation of the values) for the view or parts of the view. In one embodiment, the affected view or region is treated (eg, treated) differently from the region or view that is not affected by the out-of-field radiation source, thereby contributing to the out-of-field radiation contribution. Reduce or eliminate artifacts in the final image 156 that may result. In addition, in some embodiments, by extending the identified affected area, pixels or areas adjacent to the area believed to have been affected by an out-of-field source are also treated discriminatory, i.e. It can be considered part of the area of influence. That is, regions adjacent to or in close proximity to the regions identified as affected can also be treated discriminatory to address the effects of out-of-field sources.

In one embodiment, the affected area or view can reduce the effects of out-of-field radiation sources by removing or reducing the weight (ie, penalizing) within the data set. For example, in one embodiment, a correction (block 146) is performed by subtracting the estimated value or the contribution of the out-of-field source (eg, difference data 142) from the corresponding measured view 122. A corrected view 148 can be generated. These corrected views 148 can then be reconstructed (block 150) to generate the final image 156. In other embodiments, the affected area or view is treated differently from pixels that were not determined to be in the affected area or view by modifying or adjusting the respective camera system matrix 128. be able to. Thus, the reconstruction of the final image can be done based on the original camera system matrix 128, or based on the updated or modified camera system matrix, as shown. An example of such an updated system matrix is to display the projection data with the affected pixels P in the projection data._iA matrix that reduces by a factor (eg, f &lt;1) for pixels P_iAnd voxel V_j(M_ijThe system matrix coefficient representing the relationship with) is f * M._ijIs replaced with. The coefficient f is the smallest for the most affected pixels (ie, the most affected pixels are penalized the most or are weighted the most downward), and the distance from the affected area is It can be gradually increased to 1.0 as it increases and / or enters the unaffected area.

In certain embodiments, the final image 156 can be provided with the first reconstructed image 126, for example, in a side-by-side display array. Such sequences allow reviewers to observe and compare images with or without additional processing. Similarly, in other embodiments, the examiner may be able to observe the measured view along with the identified affected area, eg, with the affected area superimposed on the view. ..

With the above in mind, in order to explain how the effects of out-of-field radiation sources can be identified and / or addressed, the following are specific examples of various embodiments and examples. Consider in more detail. For example, in one embodiment, comparison 140 can be done in the form of pixel-by-pixel subtraction of each measured view (MV) 122 and the corresponding estimated view (EV) 132, whereby one or more. A difference view (DV) is generated. That is, DV (k, x, y) = MV (k, x, y)-EV (k, x, y) (1).

In such techniques, the differential view may be formed primarily by radiation from sources outside the field of view. As a result, the statistically high pixel value in the difference view can be displayed as the affected area. That is, pixels with values higher (or lower) than the specified threshold in the delta view can be identified as corresponding to the affected area of the corresponding measured view. Instead, smoothing the affected area and / or forcing a positive value can be used to identify the affected area in the delta view.

In other embodiments, one or more thresholds can be used to identify pixels, regions or views that are classified as affected regions. For example, in one embodiment, the average pixel value in the delta view can be calculated using all available pixel values or only positive pixel values. Similarly, the statistical standard deviation can be calculated for the associated pixel sample. The appropriate threshold for the delta view is then based on the mean pixel value for the delta view and the associated standard deviation. T_AR= (a) (APV) + (b) (SD) (2) Can be calculated according to. In this formula, "T_ARIs a pixel threshold that determines whether a pixel is classified as being within the affected area, and "APV" is the average pixel for some or all of the pixels in a given difference view. The value, "a" is the weighting factor applied to the APV, "SD" is the statistical standard deviation for the multiple pixels used to calculate the APV, and "b" is the statistical standard deviation. The weighting factor applied to SD. For example, in one embodiment, T_AR= (3) (APV) (3) "A" can be set to 3 and "b" can be set to 0 so that

In another embodiment, T_AR= (1.5) (APV) + (2) (SD) (4) "A" can be set to 1.5 and "b" can be set to 2.

Pixels in the delta view that have a value greater than the determined threshold can be marked, tagged, or otherwise identified as affected (ie, affected) pixels. it can.

In certain embodiments, the pixels identified as affected pixels can be used to identify a larger affected area within each differential image. For example, if the affected pixels in a view are greater than a specified number or percentage, the view can be defined as an affected area or view. Similarly, if the number or percentage of affected pixels in the view is greater than the specified value and a cleaning mask (discussed later) is generated for the view, then the view is affected area or view. Can be defined as.

More generally, the affected pixels in each view can be segmented, taking into account pixel strength, adjacency or proximity to other affected pixels, contiguous areas, and so on. Therefore, logic that defines the affected area or segment in each view can be used. In such an embodiment, the affected area can be identified or determined to be full, have smoothed edges, continuous, a single affected area, and so on. ..

The affected area, after being defined or divided, can be expanded (ie, subjected to expansion processing) to ensure that all affected pixels are included in the affected area. In certain embodiments, the expanded affected area can partially overlap the projection of the organ or area of interest. Such overlap may be desirable when the count within the region covered by the projection of the organ of interest is so high that it is not necessary to distinguish the affected pixels by statistical thresholds. That is, the effects of out-of-field sources can be masked or hidden by image data associated with the organ of interest.

As an example of the concept, illustrated with reference to FIG. 13, a view 180 is shown that depicts both the organ of interest (in this example, the heart 64) and an out-of-field radiation source. The out-of-field radiation source initially identifies a large number of affected pixels 182 that extend globally around the area affected by the out-of-field source (ie, the affected area). Can be distinguished. Due to threshold effects, noise, and general measurement volatility, those affected pixels are described as continuous or smooth areas, especially if the affected area and the organ of interest overlap. It is not possible. Therefore, the affected pixel 182 can be used as a base for the partitioning process, and the segmented affected area 184 is first determined by the partitioning process. The partitioned affected areas 184 can be continuous and the edges can be smoothed based on the assumptions made as part of the partitioning process. Further, in the illustrated example, the partitioned affected area 184 is expanded to generate a first extended affected area 186 that includes pixels adjacent or adjacent to the partitioned affected area 184. Can be done. In this aspect, all affected pixels are likely to be contained within the first extended affected area 186.

In the illustrated example, a second expansion can also be performed in the overlapping area between the affected area and the organ of interest to produce a second expanded affected area 190. More specifically, due to the overlap between the organ of interest and the affected area, the strong signal associated with the organ of interest causes the effects of out-of-field radiation sources in the vicinity of the organ of interest. Uncertainty (ambiguity) about existence can be greater. In the second extension, considering the overlap between the affected area and the organ of interest, the certainty that all affected pixels are covered by the area considered to be affected by the out-of-field source. To increase. For example, extension 190 may be determined based on the need for the affected area to be convex or have a limited curvature, or to match the projection of the organ from a known atlas. it can. The effects of the affected area can then be addressed as described herein, even in areas that overlap the organ or area of interest.

In certain embodiments, a cleaning operation is performed in which the organ or region of interest in the first image 126 is separated as previously described and then one or more masks derived based on the organ or region of interest are incorporated. Can be carried out. For example, in one such embodiment, the organ or region of interest has a statistically relatively high pixel value so that it can be clearly observed in the predicted view 132. The projection of the organ or region of interest on each view can be used as a mask, and the pixels outside the mask region within each view can be set to zero or background level. In certain embodiments, the pixels or regions associated with the organ or region of interest are before determining the masked region within each view to prevent any of the related pixels from being accidentally erased by masking. Can be extended to.

By voluntary choice, a composite object can be generated by constructing a 3D image with all voxels set to zero, except for voxels within a volume that is presumed to be occupied by the organ or region of interest. .. Not surprisingly, this volume can be expanded to include adjacent or other adjacent voxels to ensure that it contains the organ or region of interest. The composite object can be reprojected to create a mask for each desired view. The mask thus generated can be utilized as described above to clean (clarify) each view. Of course, in certain embodiments a cleaning operation (such as a mask-based cleaning operation) can be performed, but in other embodiments no such cleaning operation can be performed. is there.

Various techniques can be used for correcting pixel values in the affected area and / or for such differential processing of pixels. In one embodiment, pixel values within the affected area can be replaced with other values, such as estimated view 132 or values derived using a suitable model. In certain such embodiments, the entire view including the affected area can be replaced with the corresponding estimated view 132. In another embodiment, the pixel value in the affected area is the value of the affected pixel: Affected Pixel Value = (a) (MV) + (1-a) (EV) (5) It can be replaced with the weighted mean of the corresponding measured views 122 and estimated views 132, as if set according to. In this equation, 0 &lt;a &lt;1, where "a" is the appropriate weight, "MV" is the corresponding pixel value in the corresponding measured view, and "EV" is the corresponding. The corresponding pixel value in the estimated view. In other embodiments, the affected area pixel values in the delta view can be smoothed, after which these pixel values can be subtracted from the corresponding pixel values in the corresponding measured view. it can. Similarly, in another embodiment, a smoothing operation can be performed on the entire delta view and then subtracted from the corresponding measured view containing the affected area.

Further, in one embodiment, the affected pixels and the corresponding system matrix values can be replaced or modified by suppression factors (eg, downward weighted values or penalty values). For example, the pixel value can be reduced by a factor corresponding to the ratio of the "estimated in-field" count to the "estimated out-of-field" count. The system matrix elements can then be scaled in the same ratio to maintain the correct weighting. Such techniques would reduce the impact of these views without completely discarding them and without replacing the actual data with estimates.

In an embodiment where the measured view is corrected, such as setting the pixel value in the affected area as the background value to generate the corrected view 148, the final reconstruction 150 is a filter-corrected backprojection. Alternatively, it can be a standard reconstruction method based on an appropriate successive approximation reconstruction algorithm.

In other embodiments, such as those shown in FIG. 14, discriminatory treatment of the identified affected areas can be performed. For example, in one embodiment, the identified affected areas are treated discriminatively (block 200) by adjusting (eg, reducing) the values in the camera system matrix 128, resulting in a final reconstruction 150. The corrected camera system matrix 202 used in the above can be generated. As an example, the values in the camera system matrix corresponding to the affected pixels are: Corrected CSM value = (a) (first CSM value) (6) Can be replaced as Where 0 &lt;a &lt;1. As will be appreciated, the "a" can be chosen to correspond to a factor for adjusting the relational pixel values associated with the system matrix. That is, "a" can correspond to a magnification or weight that penalizes the corresponding pixel (s) due to the identified contribution from an out-of-field source. In this aspect, the relationship within the system matrix between the detector response and the observed voxel intensity can be maintained, taking into account the contribution of out-of-field radiation.

The technical effects of the present invention include the generation of reconstructed volumes with reduced or eliminated effects of emissions from out-of-field sources. The technical effect is also to compare the measured view with the estimated view to generate information about the differences between the views that represent the emission of out-of-field radiation that can cause artifacts. be able to. Based on this identification of the difference between the measured view and the estimated view, it is possible to generate a final image with no or reduced artifacts. In some embodiments, it is possible to generate a corrected view based on the identified differences. In other embodiments, discriminatory processing can be performed on the affected view region and the other view regions identified based on the above differences.

The present invention includes the present invention, including the best embodiments, to disclose the present invention, and to those skilled in the art to create and use any device or system to carry out any adopted method. Various examples were used to be able to carry out. The patentable scope of the present invention is set forth in the description of "Claims" and may include other examples conceivable to those skilled in the art. Other such examples are when they have structural elements that are substantially unchanged from the literal description of the claims, or they are substantially different from the literal description of the "claims". If it contains equivalent structural elements without, it shall be within the scope of the claims.

10 SPECT Imaging System 14 Subject 16 scanner 18 Patient support 20 Tissue or anatomy of interest 22 Detector structure 24 Collimator and scintillator 26 Photomultiplier tube 28 System control and processing circuit 60 pinhole camera 62 field of view 64 heart 66 Reconstructed volume 70 liver 72, 74, 76, 78 Heart image 90 Collimated detector assembly 96, 98, 100, 102 Heart image 180 views 182 Affected pixels 184 Divided affected areas 186 First extended affected area 190 Second extended affected area