System and method for performing non-visual diagnosis using artificial intelligence with multiple aperture ultrasound imaging

Abstract

A diagnostic ultrasound imaging method is provided. The method may include: transmitting an omnidirectional, unfocused ultrasound waveform to a target region at the transmitting aperture of a multi-aperture ultrasound imaging probe; receiving ultrasound echoes from the target region at multiple receiving apertures of the multi-aperture ultrasound imaging probe; extracting echo data from the received ultrasound echoes for a tissue density sampling region within the target region, using a processor of the multi-aperture ultrasound imaging probe, including at least one of the frequency, voltage level, and acoustic impedance for a pixel or voxel in the tissue density sampling region; identifying one or more patterns in the echo data using the processor; and correlating the one or more patterns with previously collected ultrasound echo data to determine the probability that a pixel or voxel in the tissue density sampling region contains a particular tissue type.

Description

Technical Field

[0001] (Priority Claim) This patent application claims priority to U.S. Provisional Patent Application No. 63 / 582,683, filed on September 14, 2023, entitled "SYSTEMS AND METHODS FOR CONDUCTING VISUAL AND NON-VISUAL DIAGNOSTICS USING ARTIFICIAL INTELLIGENCE WITH MULTIPLE APERTURE ULTRASOUND IMAGING", which is hereby incorporated by reference in its entirety herein.

[0002] (Incorporation by Reference) Unless otherwise expressly stated herein, all patents, publications, and patent applications mentioned in this specification are hereby incorporated by reference in their entirety, as if each individual publication and patent application was specifically and individually indicated to be incorporated by reference herein.

[0003] (Related Patents and Applications) This application also relates to the following U.S. Patent Applications: U.S. Patent No. 8,007,439 (title: "Method and Apparatus to Produce Ultrasonic Images Using Multiple Apertures"); U.S. Patent No. 9,247,926 (title: "Multiple Aperture Medical Ultrasound Transducers"); U.S. Patent No. 9,146,313 (title: "Point Source Transmission and Speed-of-Sound Correction Using Multi-Aperture Ultrasound Imaging"); U.S. Patent No. 9,668,714 (title: "Systems and Methods for Improving Ultrasound Image Quality by Applying Weighting Factors"); U.S. Patent No. 9,883,848 (title: "Ultrasound Imaging Using Apparent Point-Source Transducers") Transmit Transducer); U.S. Patent No. 10,401,493 (title: "Network-Based Ultrasound Imaging System"); and U.S. Patent Application Publication No. 2017 / 0209121 (title: "Ultrasound Imaging with Sparse Arrays"); International Patent Application No. PCT / US2020 / 05662 (title: "Systems and Methods for Tissue Characterization Using Multiple Aperture Ultrasound");This relates to International Patent Application No. PCT / US2021 / 57725 (title: "Systems and Methods for Improving Ultrasound Image Quality"), International Patent Application No. PCT / US2023 / 062069 (title: "Synthetic Aperture Unfocused Transmission for Multiple Aperture Ultrasound Systems"), and International Patent Application No. PCT / US2013 / 027120 (title: "Determining Material Stiffness Using Multiple Aperture Ultrasound"). The entire content of each of these patents and patent applications is incorporated herein by reference. These related applications may also refer herein to the prior patents and patent applications of the "Applicant."

[0004] (Field) This invention generally relates to medical diagnosis, and more particularly to diagnostic data collected through the use of ultrasound. [Background technology]

[0005] (background) Today's medical imaging utilizes multiple modalities and often requires a high level of expertise from both technicians and / or physicians to capture image data and make diagnoses. A similar level of expertise is needed to perform interventional procedures that require the placement of instruments and treatments under image guidance.

[0006] While separate from diagnostic ultrasound imaging, a related fact is that understanding physiological conditions and confirming diagnoses can sometimes be concluded through quantitative analysis of the tissue being examined, without visualizing the tissue itself. Providing qualitative and quantitative assessments of a patient's condition has become accepted in all aspects, from determining bladder volume for catheter insertion to determining gastric volume before anesthesia in high-risk patients. Perhaps best known is the non-imaging assessment of bone density using ultrasound or X-ray. In these types of cases, diagnostic data points are acceptable as a result of medical ultrasound examination, even if no images are generated. For example, an ultrasound bone densitometer can generate predictive values ​​for the t-score and z-score of osteoporosis.

[0007] Many reasons why ultrasound imaging devices are not used to provide more qualitative or quantitative data include the fact that trained technicians must direct the device to the appropriate tissue being evaluated, which usually requires imaging expertise on the part of the user. Furthermore, most conventional ultrasound imaging systems cannot image beyond conventional barriers that obstruct the transmission of ultrasound waves, such as bone and gas. These natural barriers are caused not only by signal attenuation due to large acoustic impedance mismatches between tissue and bone, but also by changes in sound velocity. For example, when imaging the liver from behind the ribs, the transmitted energy is greatly attenuated by the ribs, so only the portion of the transducer that extends beyond the ribs into the intercostal space can yield a decipherable image.

[0008] The creation of Ping-based multiple aperture imaging (PMA) has brought about significant improvements in the field of ultrasound imaging, examples of which are shown and described in the applicant's prior patents and applications. PMA is sometimes referred to as computed tomography (CET) in the prior art. Multiple aperture imaging methods and systems (PMA and CET) enable both the transmission and reception of ultrasound signals through separate apertures, either electronically or physically. This data capture method makes it possible to collect data from around conventional barriers. The resulting dataset includes multiple tissue types collected simultaneously. The dataset may also include solids, gases, liquids, and devices. For example, in cardiology, the dataset may consist of the heart, ribs, artificial heart valves (if present), moving blood, and the entire pleural cavity including the upper lungs, all of which are achievable with a PMA system. In many cases, physiological data can be collected simultaneously from multiple aperture probes and systems of multiple tissue types. The entire dataset can be viewed and analyzed as a whole, by subsection, or at the data level only without image transformation. A complete dataset of all tissue types greatly expands the ability to perform qualitative and quantitative diagnostic assessments.

[0009] This initiative concerns the collection and use of data enabled by the PMA system. Furthermore, it touches upon the use of that raw data for qualitative and quantitative analysis, as well as the use of algorithms for both learning and evaluating that data. [Overview of the project]

[0010] (Summary of Disclosure) A diagnostic ultrasound imaging method is provided. The method includes the steps of: transmitting an omnidirectional, unfocused ultrasound waveform to a target region at the transmitting aperture of a multi-aperture ultrasound imaging probe; receiving ultrasound echoes from the target region at multiple receiving apertures of the multi-aperture ultrasound imaging probe; extracting echo data from the received ultrasound echoes for a tissue density sampling region within the target region, using a processor of the multi-aperture ultrasound imaging probe, including at least one of the frequency, voltage level, and acoustic impedance for a pixel or voxel in the tissue density sampling region; identifying one or more patterns in the echo data using the processor; correlating the one or more patterns with previously collected ultrasound echo data using the processor to determine the probability that a pixel or voxel in the tissue density sampling region contains a particular tissue type; and presenting the tissue type to the user on a display device.

[0011] In some embodiments, the method includes inputting the echo data into a trained machine learning model before identifying the one or more patterns in the echo data, the trained machine learning model being configured to identify the one or more patterns in the echo data.

[0012] In some embodiments, the trained machine learning model is configured to perform the correlation step.

[0013] In one embodiment, correlating the one or more patterns further includes comparing the one or more patterns with previously collected ultrasound echo data stored in a database.

[0014] A diagnostic ultrasound imaging method is provided. The method includes the steps of: transmitting an omnidirectional, unfocused ultrasound waveform to a target region at a transmitting aperture of a multi-aperture ultrasound imaging probe; receiving ultrasound echoes from the target region at multiple receiving apertures of the multi-aperture ultrasound imaging probe; extracting echo data from the ultrasound echoes received for a tissue density sampling region within the target region, using a processor of the multi-aperture ultrasound imaging probe, including at least one of the frequency, voltage level, and acoustic impedance for a pixel or voxel in the tissue density sampling region; inputting the echo data into a trained machine learning model configured to identify one or more patterns in the echo data and correlate the patterns with known tissue types; receiving output from the trained machine learning model that identifies the tissue type of a pixel or voxel in the tissue density sampling region; and presenting the tissue type to a user on a display device.

[0015] In some embodiments, the tissue density sampling area is received as user input.

[0016] A diagnostic ultrasound imaging method is provided. The method includes the steps of: transmitting an omnidirectional, unfocused ultrasound waveform to a target region at a transmitting aperture of a multi-aperture ultrasound imaging probe; receiving ultrasound echoes from the target region at multiple receiving apertures of the multi-aperture ultrasound imaging probe; receiving input from a user indicating a tissue density sampling area within the target region; extracting echo data from the received ultrasound echoes using a processor of the multi-aperture ultrasound imaging probe, including at least one of frequency, voltage level, and acoustic impedance for pixels or voxels in the tissue density sampling area; inputting the echo data into a trained machine learning model configured to identify one or more patterns in the echo data and correlate the patterns with known tissue types; receiving output from the trained machine learning model identifying the tissue type of the pixels or voxels in the tissue density sampling area; and presenting the tissue type to a user on a display device.

[0017] In some embodiments, the method includes inputting the echo data into a trained machine learning model before identifying the one or more patterns in the echo data, the trained machine learning model being configured to identify the one or more patterns in the echo data.

[0018] In some embodiments, the trained machine learning model is configured to perform the correlation step.

[0019] In one embodiment, correlating the one or more patterns further includes comparing the one or more patterns with previously collected ultrasound echo data stored in a database.

[0020] An ultrasound imaging system is provided. The system comprises: at least one transmitting transducer element configured to transmit an omnidirectional, unfocused ultrasound waveform to a target region; a plurality of receiving transducer elements configured to receive ultrasound echoes from the target region; a processor configured to extract echo data from the received ultrasound echoes for a tissue density sampling region within the target region, wherein the echo data includes at least one of frequency, voltage level, and acoustic impedance for pixels or voxels in the tissue density sampling region, and the processor is further configured to identify one or more patterns in the echo data, and correlate the one or more patterns with previously collected ultrasound echo data to identify a probability that a pixel or voxel in the tissue density sampling region contains a specific tissue type.

[0021] In some embodiments, the system includes a display device configured to present the tissue type to a user on the display device.

[0022] In one embodiment, the processor is further configured to input the echo data into a trained machine learning model prior to identifying the one or more patterns in the echo data, and the trained machine learning model is configured to identify the one or more patterns in the echo data.

[0023] In other embodiments, the trained machine learning model is configured to correlate the one or more patterns with previously collected ultrasound echo data to identify the probability that a pixel or voxel in the tissue density sampling region contains a specific tissue type.

[0024] In some embodiments, the processor is further configured to compare the one or more patterns with previously collected ultrasound echo data stored in a database.

[0025] An ultrasound system is provided. The system comprises: a transmit aperture including at least one transducer element configured to transmit an omnidirectional, unfocused ultrasound waveform to a target region; a plurality of receive apertures each including at least one transducer element configured to receive ultrasound echoes from the target region; a processor configured to extract echo data from the received ultrasound echoes for a tissue density sampling region within the target region, wherein the echo data includes at least one of frequency, voltage level, and acoustic impedance for pixels or voxels in the tissue density sampling region, the processor is further configured to input the echo data into a trained machine learning model, the trained machine learning model is configured to identify one or more patterns in the echo data and correlate the patterns with known tissue types, and the processor is also configured to receive an output from the trained machine learning model that identifies the tissue type of the pixels or voxels in the tissue density sampling region.

[0026] In some aspects, the system includes a display device configured to present the tissue type to a user on the display device.

[0027] In one aspect, the system includes a user input device configured to receive the tissue density sampling region as a user input. [Brief Description of the Drawings]

[0028] (Brief Description of the Drawings) The novel features of the present invention are particularly set forth in the appended claims. The features and advantages of the present invention will be better understood with reference to the following detailed description that describes exemplary embodiments in which the principles of the present invention are utilized, and the accompanying drawings described below.

[0029] [Figure 1]Figure 1 is a schematic diagram of a multi-aperture imaging probe with three transducer arrays and several points to be imaged.

[0030] [Figure 2] Figure 2 is a schematic diagram of one embodiment of a concave curve transducer, showing how the transmit and receive apertures can be widened around a desired field of view to achieve higher resolution in the target region.

[0031] [Figure 2A] Figure 2A is a schematic diagram of one embodiment of a concave transducer in which the transmitting aperture and multiple receiving apertures can be electronically controlled to operate at different positions.

[0032] [Figure 2B] Figure 2B is a schematic diagram of one embodiment of a concave transducer that can quickly and continuously utilize the elements of the transmitting and receiving apertures to collect more data from multiple transmitting and receiving apertures.

[0033] [Figure 2C] Figure 2C is a schematic diagram of one embodiment of a concave transducer, showing how the transmit and receive apertures can be widened around a desired field of view to achieve higher resolution in the target region.

[0034] [Figure 2D] Figure 2D is a schematic diagram of one embodiment of a concave curve matrix having curvature in two orthogonal directions, also known as a three-dimensional (3D) array. Each element in the 3D array is displaced relative to adjacent elements in all x, y, and z axes. In this figure, one or more elements of the transmitting aperture are designated to strike the medium with ultrasound. Multiple targets in the medium are shown to illustrate how volume data can be collected. Multiple receiving apertures are shown to illustrate how timing and sound velocity adjustments within the tissue can be involved in simultaneous data acquisition.

[0035] [Figure 2E] Figure 2E shows one embodiment of a 3D array. Multiple transmit apertures T1-TN are shown to illustrate transmit pulses received at one or more receive apertures R2 and / or R3. A single target is shown to illustrate how data can be collected.

[0036] [Figure 3] Figure 3 shows a 2D grid of pixels.

[0037] [Figure 3A] Figure 3A shows an example of pixels used in a low-resolution imaging system to define points A and B.

[0038] [Figure 3B] Figure 3B shows an example of pixels used in a high-resolution imaging system to define points A and B.

[0039] [Figure 3C] Figure 3C shows an example of pixels used in a high-resolution imaging system, where individual pixels are invisible to the human eye but are clearly visible as clusters of pixels.

[0040] [Figure 4] Figure 4 shows an example of a data acquisition method using multiple aperture imaging.

[0041] [Figure 5] Figure 5 shows a raster map plotted with values ​​from different frequencies for different samples.

[0042] [Figure 6] Figure 6 shows a raster map plotted with values ​​from different samples or values ​​from signal amplitude.

[0043] [Figure 7] Figure 7 shows a raster map plotted with values ​​from acoustic impedance for different samples.

[0044] [Figure 8] Figure 8 shows an example of a data acquisition method using multiple aperture imaging. [Modes for carrying out the invention]

[0045] (Detailed explanation) Various embodiments will be described in detail with reference to the accompanying drawings. References to specific examples and embodiments are illustrative and not intended to limit the scope of the invention or the claims.

[0046] This disclosure provides a system and method that utilizes artificial intelligence to analyze raw data generated by a ping-based multiple aperture imaging ("PMA") imaging system. The obtained data can be used to improve image quality and to perform quantitative or qualitative analysis of both visual and non-visual diagnostic conditions.

[0047] As described below and in further detail in the applicant's prior patents and patent applications, PMA imaging involves transmitting an ultrasound "ping," the echo of which can be received by a receiving element located some distance from the transmitter. Each received echo signal lies along an ellipse defined by the positions of the transmitter and receiver, and the time interval between the ping transmission and the echo reception. Data elements can be created almost anywhere in a 2D grid or a 3D volume. Typically, intersecting ellipses in a 2D grid are known as pixels. Intersecting ellipses in a 3D volume are known as voxels. Images can be formed by combining such pixels or voxels (i.e., data elements) such that their intersections are highlighted. Each intersection may be an image point in a two-dimensional or three-dimensional image. Simultaneously and separately, groupings of data elements, known as datasets, can be evaluated for quantitative and / or qualitative indicators of diagnostic conditions.

[0048] Figure 1 shows a ping-based multiple aperture probe 100 S with arrays 12, 14, and 16. Subarrays within each array, or often individual elements, are shown as points a, b, c, d, e, f, g, h, and i. However, subarrays may be located across physical gaps between arrays and should not be limited to individual elements on individual arrays. Ping transmissions are represented by wavefronts 13 (dashed wavefronts) generated by the transmitting aperture at "a" on array 12, and are shown by wavelets. Point A in the medium or tissue 20 is intended to represent a hard structure (e.g., calcium or hardened plaque due to atherosclerosis, or other hard objects such as bone), where the transmitted wavefront 13 is immediately reflected or scattered in multiple directions, and is shown here as reflected wavefronts 15 (solid wavefronts). The reflected wavefronts emanating from point A can provide relatively strong (bright) signals to the receiving elements in arrays 12, 14, and 16. The transmitted wavefront 13 may also continue to propagate through the medium or tissue 20 to point B, which is intended to represent an anechoic structure (e.g., blood vessels or other soft tissue) that provides relatively weak reflected wavefronts 17 returning to the individual transducer elements on arrays 12, 14, and 16.

[0049] Echoes from points A and B can be received by receiving elements in arrays 12, 14, and 16, and can be used by probe 100 to create datasets and frames to form multi-aperture ultrasound images or for analysis by an artificial intelligence engine. One or more electronic controllers or processors associated with any probe in the PMA-enabled system can initiate the process of analyzing data in the target region. The data collected after analog-to-digital conversion is stored in a data string for a first receiving element. This receiving element may be part of an array or may be a standalone element used as an omnidirectional receiver. It does not need to be used in combination with other elements to collect and synthesize data in real time. A second receiving element can be used to generate a second string of data resulting from the same ping used to provide receiving element data to the first receiving element. Similarly, echo data resulting from the same ping transmission can be used by multiple receiving elements (e.g., third, fourth, fifth receiving elements, etc.).

[0050] In some embodiments, the transducers may be sequential, or in other embodiments, they may be configured using multiple arrays. Regardless of the type of array, the transducer elements operate independently and can be used to form any number of transmit or receive openings within the structure. When used in the embodiment of Figure 1, the terms “transducer array” or “array” generally refer to a collection of transducer elements mounted on a common backing plate. Such arrays may have one dimension (1D), two dimensions (2D), 1.5 dimensions (1.5D), or three dimensions (3D). Other dimensions of arrays may also be used, as will be understood by those skilled in the art. A transducer array may be a collection of pMUT or cMUT transducer elements. The elements of a transducer array may be the smallest independently functioning component of the array. For example, in the case of an array of piezoelectric transducer elements, each element may be a single piezoelectric crystal or a single machined section of a piezoelectric crystal.

[0051] In the embodiment shown in Figure 2, the data collected at the receiving aperture R1 comes from a single transmitting source T1 and therefore has a single, consistent speckle noise pattern. This in itself is the same limiting factor as in conventional ultrasonic systems today. However, data from the transmit T1 can also be collected simultaneously at aperture R2. The collected channel data has a different aspect angle with respect to the target and, when combined with the aperture data at R1, provides higher resolution data acquisition for pixels at the target location. In either case, energy is transmitted toward a target region having at least one reflector 170. The receiving apertures R1 and R2 can be electronically designated by MAUI electronics 140 to collect data for this transmission cycle. The subject of how to combine pixel data from receivers located along different physical paths at different apertures, and sometimes on entirely different arrays, is the subject of U.S. Patent No. 9,146,313 (title: "Point Source Transmission and Sound Velocity Correction Using Multiple Aperture Ultrasonic Imaging").

[0052] In the embodiment shown in Figure 2A, the transmitter T1 is currently located in different, physically separated apertures. Similar to Figure 2, this sequence transmitted from T1 creates a speckle pattern specific to the data received by the transducer(s) of apertures R1 and R2. However, when the dataset created in the first time in Figure 2 is combined with the data collected in the second time in Figure 2A, the aggregated dataset benefits from the constructive interference of the speckle noise patterns emanating from the different transmission locations. The clarity of the data, particularly with respect to signal frequency and amplitude, is significantly improved compared to conventional ultrasonic systems that collect data from only a single transmitting aperture or origin. Similarly, energy is transmitted toward a target region having at least one reflector 170. The receiving apertures R1 and R2 can then be electronically designated by MAUI electronics 140 to collect data for this transmission cycle.

[0053] The embodiment shown in Figure 2B is an extension of the concept shown in Figure 2A. In Figure 2B, multiple receiving apertures R1, R2, and R3 collect data from T1 in a first time. Subsequently, all receiving apertures R1, R2, and R3 receive additional data from a transmission at T2 in a second time, and additional data from a transmission at T3 in a third time. The natural advantage of this process is that it forms wider apertures and improves resolution. Figure 2B further illustrates another unique feature of some embodiments of the multiple aperture array, called field of view control. As seen in Figure 2B, the field of view can be defined as the angle α between line 180, which can be drawn from T1 to 170 along line 180, and line 190, which can be drawn from R1 to 170 along line 190. In some embodiments, the MAUI electronics 140 may be configured to automatically move the transmitting T1 aperture and the receiving R1 aperture along the array without changing the total distance between the transmitting T1 aperture and the receiving R1 aperture. This process of improving lateral resolution using a PMA system is described in U.S. Patent No. 8,007,439 (title: "Method and Apparatus for Forming Ultrasonic Images Using Multiple Apertures"). The curvature of the concave probe improves both lateral resolution (resolution measured obliquely from the center of the transducer) and axial resolution (resolution measured perpendicularly from the center of the transducer). These improvements are partly due to a reduction in the "field of view" associated with the angular offset by the receiving element, and partly due to being physically closer to the image target. Furthermore, interlacing different transmission frequencies with different transmit and receive apertures can be used to create even higher resolution datasets.

[0054] Figure 2C illustrates another important feature of several embodiments of multiple aperture arrays, called total aperture size control. In this embodiment, T 1A and R 1A The field of view created by provides an unobstructed view of the target area, including the reflector 170, avoiding the obstacle 150. In this example, the system is A 1AMultiple aperture imaging is provided with a total aperture width. Total aperture size control allows the total aperture width to be changed inward or outward on the array while maintaining a fixed field of view center. Thus, in some embodiments, as the total aperture size is adjusted, both the transmitting and receiving apertures can be electronically moved outward or inward at the same speed from the fixed field of view center while maintaining the original field of view.

[0055] Radian resolution is proportional to λ / d, where λ is the wavelength and d is the total aperture width. A wider total aperture results in higher resolution, while a smaller total aperture results in lower resolution. The total aperture width can be changed to obtain the optimal resolution for the selected viewing angle. For example, in the embodiment shown in Figure 2C, T 1B and R 1B By selecting the total opening width between A, the total opening width is maximized. 1B The total opening width can be obtained.

[0056] U.S. Patent No. 9,220,478 (title: “Concave Ultrasonic Transducers and 3D Arrays”) describes many variations of array configurations for generating volumetric datasets and their associated voxels. The diagram in Figure 2D is shown to illustrate a 3D array used with a PMA system to create a 3D volumetric dataset. Figure 2D depicts snapshots of multiple aperture data acquisition as a process of constructing an image of the entire volume 310. As seen in Figure 2D, the concave 3D transducer array 300 may have curvature around two orthogonal axes. In some embodiments, the 3D concave curve array 300 may be constructed using machined piezoelectric transducers. Because each element of the array 300 has a slightly different position on the x-axis 301, y-axis 302, and z-axis 303, calibration of the 3D array may be required.

[0057] As can be seen further in Figure 2D, snapshots of multiple aperture data acquisition are depicted as part of the process of constructing an image of the entire volume 310. Here, one or more elements of the transmitting aperture T1 transmit pulses into the volume containing scatterers such as 321 and 332. The elements constituting the receiving aperture R2 can be assembled in various shapes. Here, a square of elements constitutes the receiving apertures R2 and R3. As mentioned above, the speed of sound along the path from the transmitting aperture T1 to the reflector 321 or 332 does not affect the coherent summation of the received signals as long as a single aperture is used for reception. However, when multiple receiving apertures R1 and R2 are used, the speed of sound can be corrected to improve image quality.

[0058] In some embodiments, the size of the receiving aperture R2 may be similar to that of a conventional phased array (e.g., about 2 cm). However, unlike conventional arrays, the total aperture 340, which determines the lateral and transverse resolution of the system, is much larger because it includes the distance from the transmitter T1 to the group of receiving elements R2, and can be the same width as the entire array 300, or even wider, if the transmitter is located on a separate array within the probe (or in a separate probe communicating electronically). Each element located at the receiving aperture R2 collects volume data from the T1 transmit pulse. The size and sequence of the apertures of the transmitter and receiver can be changed throughout the array, as in the sequence described in Figure 2B, but the transmitter and receiver can now be located anywhere in the array and on any axis.

[0059] Figure 2E shows another embodiment of the 3D array 300. As seen in Figure 2E, to illustrate the transmit pulses received at one or more receive apertures R2 and / or R3, multiple transmit apertures T1~T N This is shown. A single target 321 or reflector 321 is shown to illustrate how data can be collected.

[0060] Collecting data through various tissue types naturally requires adaptation to different acoustic impedances due to varying sound velocities in each tissue. Pixel and voxel calculations can only be accurate if these adaptations are made. U.S. Patent No. 9,146,313 (titled "Point Source Transmission and Sound Velocity Correction Using Multiple Aperture Ultrasonic Imaging") details these methods.

[0061] The processor within the PMA system may be built into the probe itself or communicate with the probe (e.g., wirelessly) and be configured to perform calculations using all dataset values ​​from all data strings. In some implementations, data strings may be collected over the entire target region (i.e., a large sample period). In other implementations, data may be collected only for specific pixels and voxels (i.e., a specific sample period).

[0062] The processor can then initiate a beamforming process that generates digital values ​​at pixel locations within the target region. For 3D diagnostics and imaging, the same process can be used to generate data values ​​for each voxel. This process can be repeated for echo data strings collected from multiple ping transmissions and multiple receivers.

[0063] As shown in Figure 3, 300 represents a 2D grid of pixels. The number of pixels used in PMA can be controlled by the user. In imaging, this has the advantage of controlling the frame rate. If fewer pixels are used, processing is faster and the image frame rate can be increased. However, if the number of pixels is small, it may not be possible to produce the desired high-quality image. Therefore, the more pixels used in imaging, the more processing power is required, and the frame rate decreases accordingly.

[0064] The pixel size selected does not have to be limited to what can be visually presented. A very large number of pixels can be selected even if individual pixels are indistinguishable to the human eye. Similarly, when presenting a 3D volume of voxels, it is possible to beamform a large number of voxels that are invisible to the human eye on their own.

[0065] The number of pixels used to define points "reflectors" A and B varies and is a function of the desired resolution. Figure 3A shows a grid pattern in which pixels are located at each intersection of the X and Y axes. High-resolution systems can be more easily constructed using Ping-based multiple aperture (PMA) systems because there are no limitations on probe width, shape, or frequency. For a description of how PMA systems achieve higher resolution in ultrasound, see U.S. Patent No. 9,220,478 (title: "Concave Ultrasonic Transducers and 3D Arrays"). In such high-resolution imaging systems, pixels may be located closer to each other than in low-resolution systems. Figure 3A shows that in a low-resolution system, only a few pixels can cover point A and only a few pixels can cover to define B. Clearly, a high-resolution system has more pixels available to define both the body and edges of the reflector, thus improving the sharpness in which the reflector is presented. Figure 3B shows that more pixels are used to define the same points in a high-resolution system.

[0066] With even higher processing power, it's possible to individually define pixels and voxels that are below the visual limit. That is, at a distance of 20 inches from the display, the human eye can achieve sufficient resolution for one arcminute, which is approximately 170 dpi (dots or pixels per inch), or roughly 0.14–0.22 mm. Pixels presented below this value may be indistinguishable to the human eye. However, regardless of whether the 170 dpi requirement is met, all collected data can still be used as beamformed or beamformed data.

[0067] Figure 3C shows that in very high-resolution PMA systems, pixel values ​​can be below the limits of the visual range. Note that both points A and B have hundreds of pixels, but each is not perceptible to the human eye individually.

[0068] As a result of the beamforming process, two important pieces of data are collected for each pixel or voxel. The first is frequency. The peak frequency is determined as a function of the bandwidth selected in the system settings. A larger receiving bandwidth, selectable by the end user, widens the dynamic range, allowing each pixel to have a more accurately determined frequency value.

[0069] The second important data collected is the value or amplitude at that pixel or voxel location. The pixel or voxel value is determined by the intensity of the return signal in millivolts (e.g., the amplitude or level of the return signal).

[0070] MAUI Assigned Acoustic Impedance (MAAI) is a value generated from laboratory testing of all body tissues (e.g., bone, blood, muscle, organs, etc.) at different depths. It should be noted that MAAI is not the natural acoustic impedance of body tissue, but rather can be estimated or calculated by the system based on reflected echoes and any number of system parameters. MAAI values ​​can be stored locally on the PMA system, PMA base station, or in the cloud and are updated periodically. Acoustic impedance variations are functionally equivalent to a grayscale.

[0071] Pixel or voxel data is assigned an MAAI value by comparison with a lab-generated table. The assigned value is related to the data energy level. The higher the energy level, the higher the MAAI value. In certain embodiments, a higher MAAI value indicates a harder target tissue, allowing the data to be classified based on density.

[0072] The table below is an example of MAAI. The entries in the table are the results of laboratory characterization of data collected by the PMA system. As the artificial intelligence application (AI) is trained over time, each entry can be used as a pointer to more data. Due to the attenuation of the ultrasound signal, the data depends not only on the transducer frequency but also on the depth of the target in the body. Also note that the values ​​are unitless and are relative to the acoustic impedance of the body's soft tissues. The values ​​decrease with depth, but the ratio remains relatively constant.

[0073] Table 1: MAUI-specified acoustic impedance (MAAI) [Table 1] Methods for creating tissue density maps

[0074] To date, the methods offered have always been discussed in conjunction with the implementation of imaging. Beamforming of pixel and voxel values ​​does not necessarily require proceeding to scan conversion for image display to the user. For example, as the first step in trauma triage, it may be diagnostically desirable to determine whether there is bodily fluid inside the skull. On-site personnel may not need or be able to perform imaging. However, quantitative analysis of data performed through unguided or "blind" scans, where a probe is pressed against the target area (e.g., the skull) without imaging, can provide important physiological or diagnostic information.

[0075] As shown in Figure 4, 400 illustrates a process for achieving a tissue density that may be relevant to the diagnosis. Using a PMA system with multiple aperture probes, the probes are positioned over the area of ​​interest. The system may provide imaging data at that point according to step 401, or it may be used without providing imaging data according to step 402. The data can be collected in real time and stored in system memory, or even transmitted to and stored in the cloud during step 404 (see U.S. Patent “Network-Based Beamforming”).

[0076] Steps 406 and 408 are generated naturally as a function of beamforming in step 405. The frequency and voltage levels are then independently and grouped in steps 407 and 409, respectively. The acoustic impedance in step 410 utilizes both the frequency and voltage data of the pixel or voxel, inputting these values ​​into a known acoustic impedance table, and then in step 411, the acoustic impedance for the pixel or voxel is extracted group by group.

[0077] User controls are used to define the size of a pixel or voxel region. That is, the number of pixels to be analyzed together for pattern recognition is selectable by the user. The smallest grouping unit is a single pixel, and the largest would correspond to the system's resolution (e.g., 1260 x 1260 pixels). 302, 304, and 306 represent three pixel regions in Figure 3. Each of 302, 304, and 306 uses 6 x 3 pixels to define their region. However, a typical choice might be around 8 x 8 pixels. In a 3D system, the desired cube could be 8 x 8 x 8 voxels.

[0078] Once the region size is selected, either by the user or by a computer that automatically samples the region to search for pattern data, a raster map is created (see Figures 5, 6, and 7). These maps are visual representations of the corresponding values ​​of acoustic impedance MAAI, frequency, and voltage for that dataset at that point in time. The dataset may be a function of a single transmission (ping) and all collected channel data, or a function of multiple transmissions and their collected datasets combined. Ping-based multiple aperture imaging (PMA) does not place as much emphasis on combining data from different tissue types as conventional ultrasound systems, for reasons discussed in U.S. Patent No. 9,146,313 (title: "Point Source Transmission and Sound Velocity Correction Using Multiple Aperture Ultrasound Imaging"). Therefore, it is often possible to collect data samples from all tissue types, even though there are barriers that would normally hinder the collection of such data in conventional ultrasound systems.

[0079] In artificial intelligence or machine learning, algorithms are based on two fundamental components: (1) data, and (2) training and pattern recognition. The larger the pieces (or pixel sets) that are integrated into a recognizable pattern, the less processing is required. The smaller the pieces (e.g., below the visual limit), the more processing is required. The ability to choose the size of the blocks used to construct the baseline pattern has the natural effect of improving not only the easily recognizable data patterns but also the imaging in the same way. In other words, by making the data blocks used as small as possible, the data resolution can be improved, and consequently, the image quality can be improved. Larger data block sizes are likely to reduce the data resolution, but they can improve evaluation speed.

[0080] The second part of the process of "recognizing" patterns using AI is called training. This involves using a library of validated images, raster maps, or quantitative parameters of known tissue types. For example, the training patterns could be types of carcinoma associated with known stages of onset. A dataset of known carcinomas can be used to train a processor of many known patterns. The trained patterns are then cross-correlated with patterns collected through images, raster maps, and other quantitative data, which may provide correlation probabilities with specific pathologies (e.g., carcinoma, heart disease, vascular stenosis, etc.). The AI ​​interprets the raster maps and other data to search for patterns that match known diagnostic criteria from a library of raster maps where pathology has been confirmed.

[0081] Tissue density can be reliably estimated by correlating acoustic impedance data from the table in step 410. However, real-time analysis of tissue density is best calculated by beamforming all data in step 412. The beamforming of frequency, energy, and acoustic impedance data can then be weighted via user controls in 412. The resulting tissue density map can then be useful either within the visual range 414 or below the visual limit 416.

[0082] In some embodiments, imaging using a PMA system is performed simultaneously with beamforming of the data sample. Thus, in step 418, the imaging can be interlaced with the display of the tissue density map shown in step 420.

[0083] In systems where imaging is not performed simultaneously with beamforming of the data sample, both visual and sub-visual tissue analysis can be performed. In the case of step 416, the user can specify data regions that are clearly below the visual limit. By using these sub-visual regions together, diagnostically important conditions can be deciphered for the user, as in the case of step 422. Furthermore, AI can be used to perform sub-visual data analysis on datasets with known pathologies.

[0084] Finally, the process is repeated in real time according to step 426.

[0085] Figure 8 shows another example of a data acquisition method and system using multi-aperture imaging. In step 801, the system or method may transmit an omnidirectional, unfocused ultrasound waveform to a target region through the transmitting aperture of a multi-aperture ultrasound imaging probe. In step 802, the system or method may receive ultrasound echoes from the target region through multiple receiving apertures of the multi-aperture ultrasound imaging probe. The data may be acquired in real time, stored in system memory, or transmitted to and stored in the cloud during step 801 or step 802.

[0086] The system or method may then analyze and store the collected echo data individually or simultaneously in several ways. For example, in step 803, the system or method may analyze sample data for each pixel or voxel location based on the sample frequency of the received echo. Then, in step 806, the system or method may group the pixel and voxel data by sample frequency. Alternatively or additionally, in step 804, the system or method may analyze sample data for each pixel or voxel location according to the voltage level or voltage amplitude of the received echo. Then, in step 807, the system or method may group the pixel and voxel data by voltage level or amplitude. Alternatively or additionally, in step 805, the system or method may analyze sample data for each pixel or voxel location using acoustic impedance information. The acoustic impedance information may be measured, calculated, or derived tissue acoustic impedance values, or may be estimated or calculated by the system based on the returned echo and any number of system parameters. Here, the system or method can reliably estimate tissue density by associating the acoustic impedance information. Subsequently, in step 808, the system or method may group the pixel and voxel data by acoustic impedance.

[0087] In step 809, the system or method may beamform pixel or voxel data using weighted data from the data received from each of the aforementioned steps 806, 807, and 808. In step 809, beamforming of frequency, energy, and acoustic impedance data may be weighted automatically or via user control. Subsequently, in step 810, the system or method may identify one or more patterns in the echo data. In step 811, the system or method may correlate the one or more patterns identified in step 810 with previously collected ultrasound echo data to determine the probability that a pixel or voxel in a tissue density sampling area contains a particular tissue type. Finally, in step 812, the system or method may present or output a particular tissue type. In fact, in step 812, the system or method may present a resulting tissue density map. Such a resulting tissue density map may be either within the visual range or below the visual limit.

[0088] While the present invention is disclosed in the context of certain preferred embodiments and examples, those skilled in the art will understand that the invention extends not only to the specifically disclosed embodiments but also to other alternative embodiments and / or to the use of the invention and its obvious modifications or equivalents. Various modifications of the above embodiments will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other embodiments without departing from the spirit or scope of the invention. Accordingly, the scope of the invention disclosed herein should not be limited by the specific disclosed embodiments described above, but should be determined solely by the correct interpretation of the claims below.

[0089] In particular, materials and manufacturing techniques may be available to those skilled in the art in the relevant field. Furthermore, references to a singular item may include the possibility of multiple identical items. More specifically, the singular forms “a,” “and,” “said,” and “the” as used herein and in the appended claims include multiple referents unless the context clearly indicates otherwise. Unless otherwise explicitly stated, the term “or” as used herein includes all alternatives presented and is essentially the same as the commonly used phrase “and / or.” It should be further noted that claims may be drafted to exclude any one of their components. This statement is therefore intended to serve as a preliminary basis for the use of exclusive words such as “solely” and “only,” or “negative” limitations, in relation to the description of the components of the claims. Unless otherwise stated herein, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art in the field to which the invention pertains.

Claims

1. A diagnostic ultrasound imaging method, Transmitting omnidirectional, unfocused ultrasound waveforms to a target region through the transmitting apertures of a multi-aperture ultrasound imaging probe; Receiving ultrasound echoes from the target region at multiple receiving apertures of the multi-aperture ultrasound imaging probe; The processor of the multi-aperture ultrasound imaging probe extracts echo data from ultrasound echoes received for a tissue density sampling area within the target region, including at least one of the frequency, voltage level, and acoustic impedance for pixels or voxels in the tissue density sampling area; The processor identifies one or more patterns within the echo data; The processor correlates one or more patterns with previously collected ultrasound echo data to determine the probability that a pixel or voxel in the tissue density sampling area contains a particular tissue type; and Presenting the organization type to the user on a display device; The method comprising the step of

2. The method according to claim 1, further comprising inputting the echo data into a trained machine learning model before identifying the one or more patterns in the echo data, wherein the trained machine learning model is configured to identify the one or more patterns in the echo data.

3. The method according to claim 2, wherein the trained machine learning model is configured to perform the correlation step.

4. The method according to claim 1, wherein the correlation of the one or more patterns further comprises comparing the one or more patterns with previously collected ultrasound echo data stored in a database.

5. A diagnostic ultrasound imaging method, Transmitting omnidirectional, unfocused ultrasound waveforms to a target region through the transmitting apertures of a multi-aperture ultrasound imaging probe; Receiving ultrasound echoes from the target region at multiple receiving apertures of the multi-aperture ultrasound imaging probe; The processor of the multi-aperture ultrasound imaging probe extracts echo data from ultrasound echoes received for a tissue density sampling area within the target region, including at least one of the frequency, voltage level, and acoustic impedance for pixels or voxels in the tissue density sampling area; The processor inputs the echo data into a trained machine learning model configured to identify one or more patterns within the echo data and correlate those patterns with known tissue types; The processor receives output from the trained machine learning model that identifies the tissue type of pixels or voxels in the tissue density sampling region; and Presenting the organization type to the user on a display device; The method comprising the step of

6. The method according to claim 5, wherein the tissue density sampling region is received as user input.

7. A diagnostic ultrasound imaging method, Transmitting omnidirectional, unfocused ultrasound waveforms to a target region through the transmitting apertures of a multi-aperture ultrasound imaging probe; Receiving ultrasound echoes from the target region at multiple receiving apertures of the multi-aperture ultrasound imaging probe; Receiving input from the user indicating a tissue density sampling area within the target region; The processor of the multi-aperture ultrasound imaging probe extracts echo data from the received ultrasound echoes, including at least one of the frequency, voltage level, and acoustic impedance for pixels or voxels in the tissue density sampling area; The processor inputs the echo data into a trained machine learning model configured to identify one or more patterns within the echo data and correlate those patterns with known tissue types; The processor receives output from the trained machine learning model that identifies the tissue type of pixels or voxels in the tissue density sampling region; and Presenting the organization type to the user on a display device; The method comprising the step of

8. The system according to claim 7, further comprising inputting the echo data into a trained machine learning model before identifying the one or more patterns in the echo data, the trained machine learning model being configured to identify the one or more patterns in the echo data.

9. The method according to claim 8, wherein the trained machine learning model is configured to perform the correlation step.

10. The method according to claim 7, wherein the correlation of the one or more patterns further comprises comparing the one or more patterns with previously collected ultrasound echo data stored in a database.

11. An ultrasound imaging system: At least one transmitting transducer element configured to transmit an omnidirectional, non-focused ultrasonic waveform to a target region; Multiple receiving transducer elements configured to receive ultrasonic echoes from the target region; A processor configured to extract echo data from ultrasound echoes received for a tissue density sampling area within the target region, The echo data includes at least one of the frequency, voltage level, and acoustic impedance for a pixel or voxel in the tissue density sampling area. The processor is further configured to identify one or more patterns in the echo data and correlate the one or more patterns with previously collected ultrasound echo data to determine the probability that a pixel or voxel in the tissue density sampling area contains a particular tissue type. The system comprising the above.

12. The system according to claim 9, further comprising a display device configured to present the organization type to a user on the display device.

13. The system according to claim 11, wherein the processor is further configured to input the echo data into a trained machine learning model before identifying the one or more patterns in the echo data, and the trained machine learning model is configured to identify the one or more patterns in the echo data.

14. The system according to claim 13, wherein the trained machine learning model is configured to correlate one or more patterns with previously collected ultrasound echo data to determine the probability that a pixel or voxel in the tissue density sampling area contains a particular tissue type.

15. The method according to claim 11, wherein the processor is further configured to compare the one or more patterns with previously collected ultrasound echo data stored in a database.

16. It is an ultrasonic system: A transmitting aperture including at least one transducer element configured to transmit an omnidirectional, non-focused ultrasonic waveform to a target region; Multiple receiving apertures, each including at least one transducer element configured to receive ultrasonic echoes from the target region; A processor configured to extract echo data from ultrasound echoes received for a tissue density sampling area within the target region, The echo data includes at least one of the frequency, voltage level, and acoustic impedance for a pixel or voxel in the tissue density sampling area. The processor is further configured to input the echo data into a trained machine learning model. The trained machine learning model is configured to identify one or more patterns in the echo data and correlate those patterns with known tissue types. The processor is also configured to receive output from the trained machine learning model that identifies the tissue type of pixels or voxels in the tissue density sampling region. The system comprising the above.

17. The system according to claim 16, further comprising a display device configured to present the organization type to a user on the display device.

18. The system according to claim 16, further comprising a user input device configured to receive the tissue density sampling area as user input.

Images