Transesophageal ultrasound using a narrow probe

This application claims the priority of US Provisional Application No. 60/525330 filed on November 26, 2003.

In the medical field, monitoring of cardiac function influences critical decisions related to patient care.

One type of prior art heart monitor is an intravascular / intracardiac ultrasound transducer (such as an Accunav transducer). However, this type of transducer is not well suited for transesophageal echocardiography because the transducer elements are placed vertically rather than horizontally, limiting the types of images that can be obtained. A second type of prior art heart monitor is a transesophageal echocardiography (TEE) transducer, which is in landscape orientation. However, in order to produce reusable images, the azimuth aperture of these transducers must be extremely large (eg, 10-15 mm in diameter for adults), which requires a correspondingly large probe. To do. Due to such a large probe, conventional TEEs often require anesthesia, can significantly impair airway function, and are not well suited for long-term monitoring of the heart.<patcit num="1"><text>U.S. Pat. No. 5,417,215</text></patcit><nplcit num="1"><text>Article "Spectral Analysis of Demodulated Ultrasound Returns: Detection of Scatterer Periodicity and Application to Tissue Classification" by S. Roth, HM Hastings et al., Ultrasonic Imaging 19 (1997), pp. 266-277</text></nplcit>

<p> Transesophageal ultrasound imaging uses a small lateral transducer, preferably small enough to fit inside a 7.5 mm diameter probe, and most preferably small enough to fit inside a 5 mm diameter probe. Is implemented.</p>

<p> Signal processing technology improves penetration depth despite the fact that transducers are so small.</p>

FIG. 1 is an overall block diagram of a system that can be used to continuously monitor cardiac function through direct visualization of the heart over an extended period of time. The ultrasound system 200 is used to monitor the heart 110 of patient 100 by transmitting a drive signal to the probe 50 and processing the return signal received from the probe into the image, the image described below. Use a processing algorithm. The images produced by these algorithms are then displayed on the monitor 210 in a convenient way.

FIG. 2 shows further details of the probe 50 connected to the ultrasonic system 200. At the distal end of the probe 50 is the housing 60, and the ultrasonic transducer 10 is located within the distal end 64 of the housing 60. The next part is the flexible shaft 62, located between the distal end 64 and the handle 56. The shaft 62 must be flexible enough so that the position of the distal end 64 can reach the desired placement beyond the associated anatomy, and with the handle 56, the operator can use the distal end 64 of the distal end 64. Positioning can be done easily. Optionally, the handle 56 can include a trigger mechanism 58 that the operator uses to bend the end of the housing 60 to the desired anatomical position, as described below.

At the other end of the handle 56 is a cable 54, which terminates at the proximal end of probe 50 of connector 52. The connector 52 is used to connect the probe 50 to the ultrasonic system 200 so that the ultrasonic system 200 can operate the probe. The signal of the ultrasonic system 200 driving the transducer 10 travels through the probe 50 via proper wiring and an intermediate circuit (not shown), drives the transducer 10, and also the return signal from the transducer 10. Similarly, it passes through the probe 50 and returns to the ultrasound system 200, where it is finally processed into an image. The image is then displayed on the monitor 210 in a manner well known to those skilled in the art.

In a preferred embodiment, the outer diameter of the housing 60 is less than 7.5 mm. The probe includes an ultrasonic transducer 10 and a connecting wire, and can pass the housing 60 through the mouth or nose and into the esophagus and stomach.

The returned ultrasonic signal is processed in the ultrasonic system 200 to generate an image of the heart. As will be described later, it is preferable to further perform signal processing to greatly improve the image output. FIG. 3 shows a displayed image of a trans-gastric short axis view (TGSAV) of the left ventricle (LV), which is image processed using a preferred embodiment. It is a preferable image that can be made. The illustrated image of TGSAV is displayed in a fan shape, which includes the LV myocardium 120 surrounding the area 130 of blood within the LV. Images can be viewed in real time or recorded later for review, analysis, and comparison. Quantitative analysis of cardiac function can be performed, as appropriate, including, but not limited to, ventricular and vascular dimensions and volumes, ventricular function, blood flow, filling, pericardial structure and function, and pericardial lesions.

Unlike traditional TEE systems, the use of a relatively narrow enclosure in a preferred embodiment allows the probe to remain in place within the patient for extended periods of time.

As best shown in FIGS. 4 and 5, the probe 50 is used to introduce and position the transducer 10 at a desired location within the patient. The orientation of the heart within the thoracic cavity is such that the apex of the left ventricle is located downward and to the left. With this orientation, the lower (bottom) wall of the left ventricle is located just above the fundus, just above the left unilateral diaphragm. During operation, the transducer 10 emits a fan-shaped beam 90. By doing so, the transducer 10 is placed in the fundus and the fan-shaped beam 90 is directed upward through the left ventricle toward the heart, so that a transgastric short-axis display image of the heart 110 can be obtained. The plane of the sector beam 90 defines the image plane 95 shown in FIG. This image display is particularly useful for monitoring the work of the heart because it allows healthcare professionals to directly visualize the left ventricle and the main ventricle of the heart. In Figures 4 and 5, AO represents the aorta, IVC represents the inferior vena cava, SVC represents the superior vena cava, and PA represents the pulmonary artery. , LV represents the left ventricle.

Other transducer positions can also be used to obtain different display images of the heart, usually from the esophagus down to the stomach, allowing the operator to directly visualize most of the associated heart anatomy. can do. For example, the transducer 10 can be placed in the lower esophagus to obtain a conventional four-chamber image. Positioning of the transducer in the esophagus is usually done without completely bending the probe tip before further advancing into the stomach. Within the esophagus, have the operator use some or all combinations of probe movements, forward movements, backward movements, rotations, and light bends to obtain the desired display image of the heart.

For use by adults, the outer diameter of the housing 60 is preferably less than about 7.5 mm, more preferably less than about 6 mm, and most preferably about 5 mm. This is significantly smaller than conventional TEE probes. This size reduction reduces or eliminates the need for anesthesia and facilitates the expansion of TEE use for cardiac monitoring beyond previous dedicated short-term settings. When a 5 mm enclosure is used, the enclosure is thin enough to pass through the patient's nose, which is convenient without the risk of the patient inadvertently biting the probe. Apart from that, it can be passed through the mouth like a conventional TEE probe. The 5 mm caliber housing resembles, for example, a typical NG (naso-gastric) tube that is currently successfully used for extended periods of time without anesthesia in the same anatomical arrangement. Please note that. Therefore, it will be possible to leave the probe in place for an hour, two hours, or even six hours or more.

The housing wall is preferably made of the same material used for conventional TEE probe walls and is therefore resistant to gastric juice secretion. The wiring inside the probe that connects the transducer to the rest of the system may be similar to the wiring of a traditional TEE probe (adjusted to the number of elements, of course). The housing is preferably inserted in a relatively straight position and then maneuverable to enter the stomach and then bend to enter the proper position. The probe tip can be deflected by a variety of mechanisms, including but not limited to maneuvering or pulling wires. In other embodiments, the probe can use an intrinsic deflection mechanism, such as, but not limited to, a preformed element containing a preformed material. If appropriate, the probe (including the transducer contained therein) can also be disposable.

In order to perform image processing of TGSAV of LV, it is preferable that the probe tip is finally "bent forward" (bent forward of the patient) by about 70 to 110 degrees. This is, for example, a combination of preformed elements, a device that prevents bending during insertion, and a trigger that releases the preformed elements from insertion restrictions after the probe has been placed in the desired anatomical arrangement. It can be implemented by creating a forward bend (for example, about 70 degrees) that can be triggered through the probe. As appropriate, the draw wire can be used to steer the transducer to a further 0-40 degree bend after it has been lowered to the appropriate depth. This triggerable forward bending component is preferably designed to show little resistance to return to the unbent position when the probe is removed.

FIG. 6A shows an optional configuration similar to the embodiment of FIG. 1, except that the circuitry that interfaces with the probe 50 is relocated within the interface box 203. The rest of the ultrasonic system remains in the main processing unit 201, which communicates with the interface box 203 via the appropriate cable 205. The interface box 203 includes a circuit for amplifying and / or binarizing the signals coming from the transducer 10. Using such an interface box is convenient because it can shorten the signal path of the part of the circuit that is most susceptible to electrical noise (ie, if the signal is small). The transmitted signal driving the transducer 10 can also be generated within the interface box 203, if desired.

Electrical noise can also be reduced using a variety of techniques. For example, in one embodiment, the interface box 203 houses a preamplifier used as the first stage in the amplification / processing chain and is independent of the interface box and main processing unit 201 to reduce electrical noise pass-through. Power supply is used. In another embodiment, the interface box 203 houses a preamplifier used as the first stage in the preamplifier / processing chain, which operates on battery power. In both of these embodiments, it is preferred that time gain compensation (TGC) be implemented within the preamplifier. TGC compensates for the fact that the return signal from a distant scatterer is weaker than the return signal from a near scatterer by increasing the gain for long propagation signals. TGC can be implemented using conventional techniques well known to those of skill in the art. An example of a suitable gain-to-delay characteristic of TGC is shown in Figure 6B, where the X-axis represents the delay between the transmission of the ultrasonic pulse and the detection of the return signal, according to the following equation: Corresponds to the depth. Depth (cm) = 0.077 cm / μs x Delay (μs)

Efficient binarization can be easily achieved by implementing TGC with a preamplifier. The preamplifier can also be equipped with an amplitude expansion function (a form of compression) that facilitates efficient binarization. If appropriate, the output of the preamplifier can be binarized in the interface box, in which case only the digital signal will be transmitted from the interface box to the main processing unit, further reducing electrical noise. All possible electrical connections in the return path can be eliminated to photoisolate these digital signals and further reduce the pass-through of electrical noise.

The preferred embodiment described herein results in a high quality image of the LV TGSAV from a transducer small enough to fit within the narrow enclosure described above. 7A-7C show the first preferred transducer 10. FIG. 7A shows the placement of the transducer 10 at the distal end of the housing 60 and also includes a top view 22 and a front notch 24 of the transducer 10 surrounded by the walls of the housing 60.

As you can best see from Figure 7B, the azimuth axis (Y axis) is horizontal, the elevation axis (Z axis) is vertical, and the X axis sticks out from page to the person viewing the page. ing. Maneuvering straight forward by energizing the appropriate elements in the transducer causes the beam to exit straight along the X axis. The maneuvering signal can also emit a beam at an angle with respect to the X axis in a manner well known to those skilled in the art.

Transducer 10 is composed of N piezoelectric elements L by a method well known to those skilled in the art.₁ L_N, Acoustic backing 12, and a phased array transducer consisting of a stack of front matching layers (not shown). As one of ordinary skill in the art will understand, it is preferred that the elements of a phased array transducer can be driven individually and independently without causing excessive vibration of nearby elements due to acoustic or electrical coupling. Moreover, the performance of each element is preferably as uniform as possible in order to form a more homogeneous beam. Appointments can be incorporated into the transducer as appropriate (ie, the power-driven transducer element is tapered from the maximum in the middle to the minimum near the azimuthally end, as well as the receive gain).

Preferred transducers use the same basic operating principles as traditional TEE transducers to transmit a beam of sound energy into the patient and receive a return signal. However, while the first preferred transducer 10 shown in FIGS. 7A-7C shares many properties with conventional TEE transducers, the first preferred transducer 10 is different from conventional transducers in the following respects: different.

<tables num="1"><img file="JP4913601B2_D0001.tif" /></tables>

In FIG. 7A, the elevation angle is labeled E and the lateral opening is labeled A in the front notch 24 of the transducer 10. The wall arrangement of the housing 60 with respect to the transducer 10 can be seen in the top view 22.

FIG. 7C shows further details of the first preferred transducer 10. Although only eight elements are shown in all figures, preferred transducers actually have between about 32-40 elements, arranged at pitch P intervals, on the order of 130 μm. Two particularly preferred pitches are about 125 μm (convenient for manufacturing) and about 128 μm (0.6 wavelength at 7.2 MHz). When 32 to 40 elements are lined up at intervals of 125 μm, the resulting azimuth aperture A (sometimes simply called aperture) of the transducer 10 ranges from 4 to 5 mm. When the number of elements is reduced, the number of wires is also reduced (compared to conventional TEE transducers), which is convenient because it is easier to fit all the necessary wires in a narrower housing. The kerf K (ie, the spacing between the elements) is preferably as small as possible (eg, about 25-30 μm or less). Other preferred transducers can have elements in the range of about 24-48, spaced apart at a pitch in the range of about 100-150 μm.

The second preferred transducer 10'is shown in FIGS. 8A-8B. Transducer 10'is similar to the first preferred transducer 10 described above with respect to FIGS. 7A-7C, except that it is taller in the elevation direction. Similar reference numbers are used in both diagrams to refer to the corresponding features of both transducers. Numerically, the second transducer differs from the conventional transducer in the following points.

<tables num="2"><img file="JP4913601B2_D0002.tif" /></tables>

In another embodiment, the transducer 10 can be made to have a size that is the elevation angle of the first preferred transducer and the second preferred transducer. For example, it can have an elevation size of about 7.5 mm and a corresponding elevation: aspect ratio of about 1.5: 1.

The transducer 10 is preferably oriented sideways like a conventional TEE transducer (with respect to the axis of the housing 60). When the transducer is placed in the stomach (as shown in Figure 4), the image plane (azimuth / radial plane) generated by the transducer intersects the heart within the conventional short axis cross section. As shown in FIGS. 3 and 5, transgastric short-axis images of the heart are obtained. The transducer is preferably as wide as possible laterally within the housing. With reference to the top view 22 in FIG. 7A, two examples of transducers that fit within a 5 mm enclosure, along with a third embodiment that fits within a housing slightly larger than 5 mm, are as shown in the table below. Is.

<tables num="3"><img file="JP4913601B2_D0003.tif" /></tables>

With reference to the top view 22 of FIG. 8A, the three embodiments in Table 3 are also applicable to accommodate the second preferred transducer 10'in a 5 to 5.5 mm enclosure.

In the above embodiments, it is assumed that the housing is round. However, but not limited to, housings of other shapes, including oval, oval, etc., can also be used to house the transducer. In such cases, as used herein, when referring to the diameter of the housing, it means the diameter of the smallest circle that can surround the housing. In order to correspond to such various shapes, the housing can be specified by its outer circumference. For example, a 5 mm round housing has a circumference of 5 π mm (ie, about 16 mm). When a rectangular transducer is involved, an oval or oval housing can be used to reduce the outer circumference of the housing compared to a round housing. For example, an oval bounded by a 6 mm x 2 mm rectangle with rounded corners of 0.5 mm contains a 5 mm x 2 mm rectangular area, which holds the third exemplary transducer in Table 3. can do. By allowing a housing wall thickness of 0.04 mm, the outer circumference is 15.4 mm, which is the same outer circumference as a circle with a diameter of 4.9 mm. The table below provides the perimeter corresponding to some of the diameters described herein.

<tables num="4"><img file="JP4913601B2_D0004.tif" /></tables>

The characteristics of the last one or two elements at each end of the transducer may differ from the characteristics of the remaining elements (because their surroundings are different), so the last two elements on each side are "dummy". Can be an element. In such cases, the number of active elements driven and used for reception is the total number of elements (shown in Table 3) minus 4. Since it is not necessary for the signal to propagate to or from the dummy elements, wiring to these dummy elements can be omitted as appropriate. Apart from that, these wires can be included and can drive the last two elements, and the receive gain of those elements is partly apoded violently to correct the position of both ends of the transducer. ..

It is preferable to use conventional beam forming techniques to generate a beam of sound energy and aim it in a desired direction. For example, azimuth focusing can be done by fading (ie, individual elements L in the array).₁ L_n(Use the appropriate time delay of the returns of the individual elements before summing each return into an ultrasonic return signal). Focusing in the elevation direction can be performed based on the near-field and remote-field characteristics of the audio signal, and also depends on the elevation direction and the physical height of the optional acoustic lens element.

The resolution suitable for determining LV size and function depends on the combination of azimuth, elevation, and axis resolutions. This combination is called "spatial resolution" and is illustrated in FIG. FIG. 9 shows the image plane 320 and the scanning lines 310 on the image plane 320. Axial AX is defined by scan line 310 and the transducer (not shown) is placed far behind along the AX axis. When image processed outside the voxels, the azimuth direction AZ is perpendicular to the AX axis in the image plane 320 and the elevation axis EL is perpendicular to the image plane 320. In an ideal system, each voxel would be a point. However, in real-world systems, voxels have volumes defined by all three resolutions of AX, AZ, and EL, as shown for voxel 330. Similarly, although the image plane 320 is shown as a thin plane, the real-world image plane has a thickness in the elevation EL that is equal to the thickness of the voxel 330 in the elevation.

The general formula for azimuth and elevation resolution is: Δθ 1.22λ / d Where Δθ is the beam width in radians, λ is the wavelength (corresponding to the center frequency of the transducer), and d is the aperture in a given direction (azimuth or elevation). Wavelength λ and aperture d are measured in the same units (eg, μm).

The axial resolution is indirectly dependent on the wavelength λ. The inventor is unaware of the specific formula for axial resolution, but is usually around 16-64 times the wavelength. Therefore, increasing the center frequency increases all three components of spatial resolution. The center frequency of about 5 to 10 MHz is a high frequency that gives appropriate resolution.

FIG. 10 illustrates the interaction between the three components when examining the interaction between the resolution voxel shape and the boundary orientation when detecting and determining the boundary. It shows the same voxel 330 as shown in FIG. 9, and also shows the exemplary component 340 of the image-processed boundary that matches the voxel. If the boundary orientation is random with respect to the resolution voxels, one preferred approach is to make the resolution voxels as cubic as possible. In order to obtain that shape, the azimuth and elevation resolutions of a given box must be approximately equal, which is similar to the first preferred transducer described above with respect to FIGS. 7A-7C. Occurs when the front of the is approximately square.

In the first preferred transducer, the elevation opening is about the same as the azimuth opening. That is, the front surface of the transducer has an elevation: lateral aspect ratio of approximately 1: 1 (ie, approximately square). Therefore, a square transducer with a lateral width of 4-5 mm is about 16-25 mm.²Has an area of.

The formulas for azimuth and elevation resolutions are: Δθ_AZ= 1.22 × λ / d_AZ as well as Δθ_EL= 1.22 × λ / d_EL However, Δθ_AZAnd Δθ_ELAre azimuth resolution and elevation resolution, respectively (both measured in radians), d_AZAnd d_ELAre the azimuth opening degree and the elevation angle opening degree, respectively. These components can be combined into a single equation for the overall resolution as a function of area and frequency as follows. Δθ_OVERALL= 1.5 × λ²/ (d_AZ× d_EL)

As explained above, increasing the center frequency increases the resolution. However, increasing the center frequency further reduces the penetration depth due to frequency-dependent attenuation, to which the following approximation formula applies. α 0.5f × r Where α represents unidirectional attenuation in dB, f represents the center frequency in MHz, and r represents the depth in cm. Therefore, the one-way frequency-dependent attenuation is typically about 0.5 dBMHz.^-1cm^-1And a typical round-trip frequency-dependent attenuation is typically about 1 dBMHz.^-1cm^-1Is.

The inventor has adopted a transducer center frequency in the range of about 6 to 7.2 MHz, which is a good trade-off between resolution and penetration depth for TEE using a transducer with an azimuth aperture of 4.75 mm. Judging that a relationship can be obtained. In the embodiments described herein, the frequency range usually provides sufficient penetration depth to image the back wall of the left ventricle (in TGSAV) and the internal volume of the left ventricle. (For most subjects, a penetration depth of 12 cm is sufficient to image the back wall. For many subjects, a penetration depth of about 9-10 cm is sufficient. ).

When the transducer elements are spaced 125 μm pitch apart, using a transducer center frequency of 6.16 MHz is particularly convenient as it corresponds to a wavelength of λ = 250 μm. At that wavelength, the elements are spaced at a pitch of 0.5λ, also known as the 1/2 wavelength pitch. As is well known to those of skill in the art, the 1/2 wavelength pitch is excellent at eliminating grating lobes while minimizing the number of elements for a given azimuth aperture. Somewhat larger pitches, such as 0.6λ, work well enough with respect to eliminating grating lobes. Therefore, a transducer that can operate at a center frequency in a certain range can maintain acceptable performance even when the frequency is increased by about 20% (that is, to the point where the pitch becomes about 0.6λ).

As explained above, the formula for angular resolution is θλ / d. With reference to the table above, the first embodiment of the first preferred transducer comprises a 38 element transducer with a pitch of 125 μm, resulting in a transducer width of 4.75 mm (d = 4750 μm). It is approximately square and preferably operates at a center frequency of 6.16 MHz (λ = 250 μm). Inserting these values for d and λ into the resolution equation yields the result θ 0.053 radians, which translates to a resolution of about 3 degrees in both azimuth and elevation.

The larger size of the transducer in the elevation direction makes it easier to improve the angular resolution of the system in the elevation direction (compared to traditional TEE transducers with an elevation angle of 2 mm). When the resolution is increased in the elevation angle direction in this way, it becomes easy to correct the loss amount of the azimuth angle resolution caused by reducing the azimuth angle opening degree to about 4 to 5 mm.

The inventor has found that increasing the size of the transducer further in the elevation direction to be larger than the size in the azimuth direction improves performance when imaging the inner wall of the TGSAV heart. Such an increase in the elevation angle of the transducer causes the resolution voxels to contract in the elevation angle direction at a distance corresponding to the wall at the back of the LV, resulting in higher resolution in the elevation angle direction. The inventor is convinced that increasing the resolution in this direction is beneficial, at least in part, because the back wall is tilted around the Y axis with respect to the front of the transducer (Y axis is Figure 8B). It is shown in.). Therefore, reducing the voxel size in the elevation direction minimizes the variation in the component of the return signal resulting from specular reflection within a single voxel.

As in the case of the second preferred transducer 10'described above with respect to FIGS. 8A and 8B and Table 2, the inventor imaged the TGSAV with an elevation angle of the transducer being more than 1.5 times larger than the lateral direction. Sometimes good, we determined that the best image of TGSAV was obtained when the transducer's elevation was about twice as large as the lateral.

In a preferred embodiment, a smaller sector width (eg, 60 degrees) is used instead of the 90 ° sector width typically used in traditional TEE systems. With reference to FIG. 11, the 60 ° sector 92 is shown to extend from the front 14 of the transducer 10. The effective azimuth opening degree forming an angle θ from the center line CL is obtained by multiplying the (nominal) azimuth angle opening degree (θ = 0) by cos (θ). Since cos (30 °) = 0.866 and cos (45 °) = 0.707, limiting the width of the sector to 60 ° (that is, 30 ° on both sides of the centerline CL) reduces the azimuth opening in the worst case. Is smaller and the azimuth opening is only 13.4% lower than the 26.8% for a 90 ° sector width. For example, the worst aperture for a 4.75 mm wide transducer (5 mm housing diameter) in a 60 ° sector is about 4.11 mm. As a result, the effective azimuth aperture is improved and the overall resolution obtained with a small transducer is improved. If a traditional 90 ° sector were to be used, a 5.82 mm wide transducer (6.1 mm housing diameter) would be required to obtain the same worst-case aperture.

After a beam of ultrasonic energy is transmitted into the patient body using the transducer described above, the ultrasonic return signal is preferably received by the same transducer. The transducer converts the ultrasonic return signal into an electrical return signal. This process continues as the beam is swept within the imaging sector. Figure 12 shows scan line B first.₁Along, then scan line B₂Along and scan line B_MIt is a schematic diagram of the path of the ultrasonic beam drawn when swept in a sector, continuing through. These scan lines B₁ B_MCorresponds to sector beam 90 (shown in FIG. 4) and sector 92 (shown in FIG. 11). The figure contains only a few (M) scanlines, but in a real system there are more scanlines that are fairly dense so as not to adversely affect azimuth resolution.

The electrical return signal can be modeled as an amplitude modulated signal, the carrier frequency is at the center frequency, and the modulation is mostly caused by other tissue characteristics such as the spacing of the scatterers and the presence of connective tissue around the myocardial bundle. Is done. The electrical return signal is demodulated and binarized (ie sampled) to form a demodulated and digitized return signal (DDRS). DDRS can be formed using a variety of conventional techniques well known to those of skill in the art. In one embodiment, the electrical return signal is binarized and the result is adjusted (ie, taken in absolute value) to form a calibrated binarized ultrasonic return signal. In another embodiment, the electrical return signal is tuned in analog form and then the result is binarized to form DDRS. Extracting modulation information from an electrical return signal using an alternative demodulation approach, including, but not limited to, coherent demodulation, Hilbert transforms, and other demodulation techniques well known to those of skill in the art. You can also.

FIG. 13 shows the ultrasonic beam B of FIG.₁ B_MIt is a schematic diagram of DDRS corresponding to one part of. Each sample is represented by dots S0 ... S143. Each sample corresponds to a point in 2D space based on the direction of the beam and the time it takes for the signal to propagate from the transducer to the point of interest and return. For example, if the return signal is binarized at 50MHz, the time between samples is 0.02μs, which corresponds to a distance of 0.015mm (based on the speed of sound of the object). The figure contains only 144 samples, but in a real system, more samples are included within each scan line to obtain the desired resolution. For example, 8000 samples would be required to obtain a penetration depth of 12 cm with a sample spacing of 0.015 mm. Since the beam of ultrasonic energy is swept around the center point, polar coordinates are useful for organizing the sample, at least at this processing stage. In some embodiments, all of these samples are analyzed in polar coordinates and converted to Cartesian coordinates only when displayed on a conventional computer monitor. In other embodiments, the sample space can be converted to Cartesian coordinates early in the process. In the rest of the description, the coordinates along each scan line (the constant θ is in the (r, θ) polar coordinate system and r changes along the scan line) and the center of the pixel along that scan line. Consider the pixel data related to. Conversion to a sector image is well known in the field of ultrasound imaging.

Each scan line sample is preferably processed by two different algorithms, one that analyzes the intensity characteristics of the sample and one that analyzes the frequency characteristics of the sample.

In the first algorithm (ie, the intensity algorithm), the scan line sample is divided into multiple pixels, each pixel containing a plurality of samples. In the example of FIG. 13, each pixel is labeled "WIAP j" (short for "Window for Intensity Algorithm for Pixel j", where j is an integer from 0 to 8) displayed below the corresponding sample. Contains 16 samples, as shown in the box with. The pixel data generated by signal processing is associated with the center position of the corresponding pixel. Of course, it is possible to use a number other than 16 as the number of samples per pixel. In a preferred embodiment, for example, each pixel comprises eight samples. The intensity algorithm is preferably a conventional image processing algorithm that converts those samples into a conventional image. The intensity for a given pixel is determined based on the amplitude of the sample corresponding to that pixel, with higher intensities corresponding to larger amplitudes. For 16 sample pixels, the average of those 16 samples is used to determine the intensity at that pixel (higher average intensity values appear brighter, lower average intensity values appear darker). ). Optionally, the intensity level of the pixel (or the samples that make up the pixel) can be compressed using conventional methods such as logarithmic compression.

The second algorithm (ie, the frequency algorithm) analyzes the frequency characteristics of the sample space to determine the spatial frequency of the scatterer spacing. Examples of suitable algorithms are described in Patent Document 1 incorporated herein by reference. Non-Patent Document 1 is also incorporated herein by reference.

The frequency algorithm gives a second result for each pixel in the image (that is, in addition to the result produced by the intensity algorithm). For most frequency analysis algorithms, the more data samples used, the better the results, and since each pixel has only a limited number of samples, the samples on either side of the pixel of interest are the pixels themselves. It is preferable to increase the number of samples in combination with the samples of. In the example shown, each pixel contains 16 samples, but the "Window for Freq." Displayed below the samples. As shown by the box labeled "Algorithm for Pixel k" (k is an integer in the range 2-6), the frequency algorithm for a given pixel is preferably 64 pixels centered on that pixel. Works on the sample. For example, in this case, a frequency analysis for pixel 3 is performed using all the samples from pixels 2-4 plus half of the samples from pixels 1 and 5. Of course, it is possible to use other numbers of samples for frequency analysis instead of 64. However, a power of 2 is preferred when the Fast Fourier Transform (FFT) algorithm is used. If appropriate, windowing techniques (such as humming windows) can be used to give more weight to the central sample than to the near-end sample.

FIG. 14 is a flow chart of a suitable frequency algorithm. In this algorithm, steps 1 and 2 are collectively the substances that make up the pixel (and more specifically, the pixel is blood) of interest based on the frequency characteristics of the samples within that pixel and the samples within its neighbors. Attempts to identify (whether it is muscle or muscle).

In step 1, a Fourier analysis is performed on the sample to determine the power distribution of each pixel in different frequency bands. The final result of the Fourier analysis in step 1 is, for each pixel, a set of amplitude coefficients for several different frequencies (ie, a set of 1 coefficient for the first pixel, a second set of coefficients for the second pixel, and so on. ). Fourier analysis can be implemented using a variety of algorithms well known to those of skill in the art (eg, conventional FFT algorithms). In other embodiments, similar results can be obtained using other frequency analysis tools such as the band method (preferably integer-based FIR recursion), the wavelet method, and the like. In step 2, the ratio of the power in the selected frequency band to the power in the entire spectrum for each pixel is calculated. Therefore, the following formula is applied for each pixel. R = E_BAND/ E_TOTAL However, E_BANDIs the power in the selected frequency band, E_TOTALIs the total power in a part of the spectrum and R is the ratio of those two powers. When Fourier analysis is used, the power in a given band is equal to the sum of the squares of the amplitudes of the Fourier coefficients in that band. The "selected frequency bands" of this step are preferably selected so that the change in ratio R correlates with the difference in the material being imaged (eg, blood vs. muscle). Separately, changes in the ratio R can be chosen to correlate with the difference in signal-to-noise ratio, with large R correlating with the signal and small R with speckle or electrical noise. As appropriate, different "selected frequency bands" can be used for near and far returns. For example, a wide frequency band can be used for signals corresponding to distant structures. That is, bandwidth selection can be a function of depth.

One suitable numerical set that provides a correlation between R and the material to be imaged will be described. First, consider the ultrasonic waves returning from a single scatterer at a depth of rmm. The ultrasonic waves returning from this scatterer arrive after a delay time of t [μs] given by the following equation. t = r / ν = r / (0.77mm / μs) = 1.30r [μs] However, a magnification of 0.77 mm / μs represents a round trip from the transducer to the scatterer and then back (assuming the velocity of sound in the tissue is 1.54 mm / μs).

The effect of the scatterer's periodicity on the spectrum of the returned ultrasonic waves can be calculated if they are sufficiently spaced apart so that the returned ultrasonic waves do not overlap (that is, the interval Δr is Δr.₀= 0.77 mm / μs × greater than Δt). For example, for an ideal 1-cycle pulse, a center frequency of 5 MHz, and an ideal wideband transducer. Δt = 1 / f_c= 1 / (5MHz) = 0.200μs Therefore, Δr₀= 0.77 mm / μs × 0.200 μs = 0.154 mm.

The internal structure of the myocardium displays fluctuations on this spatial scale and on larger spatial scales. In contrast, scattering from blood is characterized by fully developed speckles, including all and especially fairly small, spatial-scale variability. As a result, low frequencies indicate muscle and high frequencies indicate blood. This suggests that the upper limit of the low frequency band is defined to be less than about 4 MHz, which is the minimum spatial scale Δr below._MINCorresponds to. Δr_MIN= 0.77 mm / μs × 1 / (4 MHz) = 0.77 mm / μs × 250 μs = 0.193 mm

The inventors conducted a tissue experiment using a signal binarized at 50 MHz (corresponding to a sampling interval of 0.02 μs) and performed a 64-point window (corresponding to 64 × 0.02 μs = 1.28 μs or 0.986 mm). The FFT was calculated within. For windows of that size, the inventor chose a low frequency band containing Fourier frequencies in the range of 2 to 5 cycles (including) per window, which was 2 / 1.28MHz = 1.56MHz and 5 / 1.28. It corresponds to frequencies in the range of MHZ = 3.91MHz.

The above R (R = E) for this low frequency band_BAND/ E_TOTAL) Is used to calculate the ratio of the Fourier power in the low frequency band to the total Fourier power pixel by pixel. The final result of step 2 in FIG. 14 is the value of R per pixel.

For the parameter values used in this example, the inventor found that an R value centered on 0.45 correlates with the presence of muscle tissue in the pixel of interest, and an R value centered on 0.20 is electronic noise. Was found to correlate significantly with the dominant blood or region. The rest of the algorithm uses this information to improve the image by increasing the intensity of the image portion corresponding to the muscle and decreasing the intensity of the image portion corresponding to the blood. This difference increases the contrast between blood and muscle, as blood has less reflexes than muscle.

The inventor found that the intensity of the region with the R value corresponding to muscle increased to about 120% of its original value, and the intensity of the region with the R value corresponding to blood increased from about 20% to 50% of its original value. When it dropped to, it was judged that the echocardiography had improved dramatically. Therefore, in step 3 of FIG. 14, a gain coefficient of about 1.2 is assigned to the part of the image with an R value of about 0.45, and a gain coefficient between about 0.2 and 0.5 is assigned to the part of the image with an R value of about 0.20. Assigned. Since this gain coefficient depends on the characteristics, it is referred to as a "feature gain factor" or FGF (feature gain factor) in the present specification.

Most pixels in most images have an R-value that allows them to be classified as either muscle or blood, but in some cases this classification may not be very clear. For example, a pixel that straddles the boundary between muscle and blood has a less predictable R value. Moreover, the R value from blood averages 0.20, but a given pixel of blood can vary significantly from that R value due to random statistical deviations. Therefore, in some embodiments, a monotonous, preferably smooth function can be used to map R to FGF. Figure 15 is an example of a function that serves this purpose. If appropriate, additional restrictions can be placed on the mapping function based on other organizational characteristics.

Finally, in step 4 of FIG. 14, the intensity algorithm result and frequency algorithm are obtained by multiplying the intensity value for each pixel (obtained from the intensity algorithm) by the FGF value for that pixel (obtained from the frequency algorithm). Combine the results. The result is a highlighted image where the pixels, which are probably blood, are darkened, while the pixels, which are probably muscles, are brightened. The highlighted image is then displayed using conventional hardware and software techniques (including, for example, using interpolation methods that convert polar coordinates to Cartesian coordinates).

The actual selection of the Fourier frequency band, R value, and corresponding FGF value is used in, but is not limited to, the transducer center frequency, sampling rate, window size, and signal processing, transducer bandwidth, and communication pulse width. It depends on a variety of factors, including optional window processing techniques. For example, in one embodiment, a transducer center frequency of 7.5 MHz is used, the scan lines are binarized at about 4 times the center frequency (ie, about 30 MHz), and the distance between samples is about 0.026 mm. ..

In other embodiments, the other normalized one (ie, the non-amplitude dependent scale) is E._BANDE_TOTALCan be used instead of dividing by. For example, using the ratio of the power in the first frequency band to the power in the second frequency band, as described in Patent Document 1 (eg, E)._BAND1E_BAND2R can be calculated (by dividing by). In other embodiments, two or more Fourier analyzes can be performed pixel by pixel using the corresponding number of scan lines in a sample whose center of each scan line is within a pixel. For example, one pixel where the first 1D Fourier analysis is performed radially along one scan line of the sample and the second 1D Fourier analysis is performed tangentially along the second scan line of the sample. It is an array of 2 scan lines per unit. The results from these two scan lines of the sample are then merged (eg, by averaging). In yet another embodiment, the 2D Fourier algorithm can be used in place of the 1D algorithm described above.

Usually, the above calculation is performed on uncompressed image data. However, under certain circumstances, it may be possible to perform the corresponding operations directly on the compressed version of the image data.

After the highlighted image is generated, it can be displayed using conventional hardware. Images can be continuously updated and displayed throughout the time the probe is in place, allowing doctors to visualize the patient's heart in real time. In other embodiments, it can be acquired periodically (eg, by capturing one or more complete heartbeats every two minutes) and stored as appropriate. As appropriate, the ability to compare the previous heart rate to the current heart rate, for example, play a video clip (or "loop") containing the previous heart rate in one window and the current image in the second window. It can be realized by displaying in.

In contrast to traditional long-term TEEs that use transducers with an azimuth opening of 10-15 mm, performed under general anesthesia in a tightly monitored environment in a room, as used herein. Due to the small diameter of the preferred embodiment described, this preferred embodiment can be used without the use of general anesthesia and in a less rigorous surveillance environment. Optionally, preferred embodiments can be used with sedation or local anesthesia instead of general anesthesia used with conventional long-term TEE. It may even be possible to avoid the use of sedation or anesthesia at all. In such cases, the patient may receive appropriate analgesic administration.

As appropriate, areas with high relevance, such as those detected by the feature gain coefficient, are highlighted by using normal colorization, while preserving the intensity of the grayscale image, as described in Patent Document 1. can do. Other techniques for image enhancement can be found in US Pat. No. 6,932,770, entitled "Method and MFP for Ultrasonic Imaging," filed August 4, 2003, which is incorporated herein by reference. ..

In the preferred embodiment described above, a small transducer that fits in a housing about 5 mm in diameter can be conveniently used for non-invasive mid- and long-term monitoring of cardiac function, and therefore the need for anesthesia is low. Or not at all. In the preferred embodiment described above, a plurality of techniques are combined to produce an image of quality comparable to or better than that previously obtained with a fairly large transducer. The images produced by the preferred embodiments described above are LV size and from an image of the endocardial wall with an appropriate penetration depth (10-12 cm) where the inner wall of the left ventricle is visible, despite the use of a small transducer. It can be repeated and reliably used to monitor cardiac function with sufficient resolution to determine function. Thus, in a preferred embodiment, in contrast to prior art systems that provide penetration depths that are less than 15 times the azimuth opening of the transducer (eg, a 10 mm transducer is used to obtain 10 cm penetration). Penetration can be achieved over 15 times the azimuth opening of the transducer, or even 20 times the azimuth opening of the transducer (eg, use a 4.75 mm transducer to get 10 cm penetration).

In the preferred embodiment described above, a probe that is considerably thinner than a conventional TEE probe is used and can be used to monitor cardiac function over an extended period of time and to understand the hemodynamics of the patient. Such information may be useful in choosing treatment and improving outcomes in many situations, including but not limited to serious medical problems such as hypotension, pulmonary edema, and heart failure. Is done.

In the embodiments described above, cardiac function can be directly visualized, so that intravascular volume (normal, low or high), myocardial contractility (condition of pumping function of the left ventricle), cardiac ischemia (blood to myocardium). It is possible to assess the hemodynamics of a patient, including (lack of flow), cardiac tamponade (fluid in the pericardial sac that limits cardiac function). For example, information about vascular volume status can be derived by directly visualizing the size of the left ventricle and monitoring changes in size over time with treatment. Information about contractility can be obtained using qualitative visual inference or quantitatively by directly visualizing the contraction of the left ventricle (pumping motion). Information on ischemia is available during direct visualization of the left ventricle because the wall of the left ventricle behaves abnormally due to ischemia (abnormal wall motion). Information about possible cardiac tamponade or epicardial fluid (fluid in the pericardial sac) is available when directly visualizing the heart using ultrasound.

Due to the thinness of the probe, the above embodiments can provide this information for extended periods of time outside the operating room and / or without the use of anesthesia. The embodiments described above are further performed by interventional cardiac therapies such as cardiac catheterization and electrophysiological experiments to monitor the effects of physician intervention on cardiac and hemodynamic function and guide the placement of the device. Useful for use in settings that are made. For example, they can be used by a physician to properly position the pacing reeds and achieve the desired results. The embodiments described above can also be used in applications other than cardiac therapy where a fine probe is required or beneficial.

The above-described embodiment is not limited to the ultrasonic imaging mode, and can be used in other ultrasonic modes (for example, pulse wave Doppler, continuous wave Doppler, and color blood flow imaging Doppler mode). These alternative modes can be performed using the same transducers as the imaging modes described above and can produce information that can be combined with the image in real time as appropriate. For example, color blood flow Doppler information can be obtained during imaging of the mitral valve (between the left atrium and the left ventricle) while maintaining the position of the transducer between the middle and lower esophagus. In such applications, mitral valve leakage (mitral regurgitation and insufficiency) could be assessed.

If desired, the preferred embodiments described above can be reduced to newborn or pediatric use. In such cases, a transducer between about 2.5 and 4 mm in the azimuth direction is preferred and the elevation dimension is proportionally reduced. Because newborn or toddler patients require a small penetration depth, the operating frequency can be increased. As a result, λ becomes smaller, the spacing (pitch) of the transducer elements used can be reduced, and the corresponding number of elements per 1 mm in the transducer can be increased. When such a transducer is combined with the techniques described above, it will meet or exceed the performance of conventional 7.5 mm TEE probes for neonatal and pediatric applications.

The embodiments described herein can also be used in applications other than cardiac therapy. For example, a probe can be inserted into the esophagus to monitor the patient's esophagus itself, lymph nodes, lungs, aorta, or other anatomical site. Alternatively, the probe can be inserted into another orifice (and even an incision) to monitor other parts of the patient's anatomy.

If necessary, the center frequency can be lowered (eg, down to about 4.5MHz) to increase the penetration depth when needed (eg, for very large patients). This further reduces the resolution, but the results are acceptable when very large structures are imaged. Alternatively, the size of the transducer and the diameter of the housing can be increased (eg, about 7 mm) if the image becomes unusable as a result of reduced resolution.

A number of alternative and optional features can be replaced and added to the embodiments described above. One optional feature is the use of significant oversampling to perform digital beam formation. For example, if the transducer operates at 7MHz and the return signal is binarized at 30x frequency, then binarization of 30x7MHz = 210MHz is required. The data could then be downsampled with a factor of 5 to reduce the number of data points to a 42 MHz sample. Such downsampling reduces the noise level due to front-end noise by a factor of 5 (that is, more than 2 bits in power). Similarly, downsampling with a factor of 7 reduces the noise level by a factor of 7.

FIG. 16A shows the front of the Alternate 2D Transducer 500, which includes a 2D array of active elements 510. The concepts described herein can also be implemented using this type of transducer with appropriate adjustments that will be apparent to those of skill in the art.

Figure 16B shows the front of another alternative 2D transducer design called a "sparse 2D transducer". The sparse 2D transducer 600 has column 610 of "transmit" element 611 used to transmit ultrasonic waves and row 620 of receiving element 621 used to receive ultrasonic signals. As shown in the figure, there is one element 630 in common in both the sending element column 610 and the receiving element row 620. This common element 630 can be used for transmission, reception, or both. This transducer design reduces electronic noise by using independent transmit and receive elements that do not require an electronic transmit / receive switch in the element. The concepts described herein can also be implemented using this type of transducer with appropriate adjustments that will be apparent to those of skill in the art.

Other embodiments of the invention can use fewer techniques and / or implement them, if not so much, and still maintain the ability to produce acceptable images. For example, depending on the other components in the system, it is possible to use a sector width of 75 ° or even a sector width of 90 ° to obtain an acceptable image. It is also possible to use a 2: 3 elevation: lateral aspect ratio transducer to obtain an acceptable image instead of the preferred 1: 1 or 2: 1 aspect ratio. Other alternatives use some or all of the techniques described above with transducers that are slightly larger than the preferred embodiments described above, yet this is still smaller than conventional 10 mm TEE transducers. Many other modifications to the above embodiments will be apparent to those skilled in the art and are also within the scope of the present invention.

<figref num="1">An entire block diagram of a system that monitors cardiac function through direct visualization of the heart.</figref><figref num="2">It is a more detailed view of the probe shown in the embodiment of FIG.</figref><figref num="3">It is a schematic diagram of the display image of the transgastric short axis image (TGSAV) of the left ventricle.</figref><figref num="4">It is a figure which shows the positioning of a transducer with respect to a heart to acquire TGSAV.</figref><figref num="5">It is a figure which shows the plane which opened in the transgastric short axis of the heart.</figref><figref num="6A">It is a figure which shows the optional probe interface configuration.</figref><figref num="6B">It is a graph of the gain characteristic of a TGC amplifier.</figref><figref num="7A">It is a figure which shows the 1st preferable transducer structure.</figref><figref num="7B">It is a figure which shows the 1st preferable transducer structure.</figref><figref num="7C">It is a figure which shows the 1st preferable transducer structure.</figref><figref num="8A">It is a figure which shows the 2nd preferable transducer structure.</figref><figref num="8B">It is a figure which shows the 2nd preferable transducer structure.</figref><figref num="9">It is a figure which shows the component of a spatial decomposition.</figref><figref num="10">It is a figure which shows the interaction between the shape of a resolution voxel and a boundary.</figref><figref num="11">It is a figure which shows the width of a sector.</figref><figref num="12">It is a schematic diagram of the path of an ultrasonic beam drawn when swept in a sector.</figref><figref num="13">It is a schematic diagram of a sample corresponding to one part of the beam of FIG.</figref><figref num="14">It is a flow chart of the processing algorithm which uses the frequency characteristic of a return signal.</figref><figref num="15">It is a graph of the function which maps the gain coefficient to the energy ratio.</figref><figref num="16A">It is a figure which shows two alternative transducer designs.</figref><figref num="16B">It is a figure which shows two alternative transducer designs.</figref>

Code description

10 Ultrasonic Transducer 12 Acoustic backing 14 Front 22 Top view 24 Front notch image 50 probe 52 connector 54 cable 56 handle 58 Trigger mechanism 60 chassis 62 shaft 64 Distal end 90 fan-shaped beam 92 60 ° fan shape 100 patients 110 heart 120 myocardium 130 Blood area 200 ultrasonic system 201 Main processing unit 203 interface box 205 cable 210 monitor 310 scan line 320 image plane 330 voxels 340 components 500 alternative 2D transducer 510 active element 600 sparse 2D transducer 610 columns 611 "Send" element 620 lines 621 Receive element 630 elements