Ultrasound diagnostic device, control method for ultrasound diagnostic device, and control program for ultrasound diagnostic device

Abstract

To provide an ultrasonic diagnostic device capable of improving accuracy of a likelihood image indicating a target region in ultrasonic image diagnosis using a spatial compound method.SOLUTION: An ultrasonic diagnostic device 1 includes: a transmission / reception unit 11 for executing transmission and reception of an ultrasonic beam for an ultrasonic probe 20; a signal processing unit 12 for generating an ultrasonic image on the basis of a reception signal acquired from the ultrasonic probe 20; a target identification unit 13c for executing segmentation processing based on a structure type for the ultrasonic image, and generating a likelihood image indicating a presence region of a target in the ultrasonic image; and a likelihood image composition unit 13e for compositing the likelihood image of each of a plurality of the ultrasonic images generated by ultrasonic scanning using ultrasonic beams of mutually different steering angles, and generating a spatial compound likelihood image.SELECTED DRAWING: Figure 8

Description

[Technical Field]

[0001] The present invention relates to an ultrasonic diagnostic apparatus, a control method for an ultrasonic diagnostic apparatus, and a control program for an ultrasonic diagnostic apparatus. [Background technology]

[0002] Conventionally, as one type of medical image diagnostic device, an ultrasound diagnostic device has been known that transmits ultrasound waves toward a subject, receives the reflected waves, and performs predetermined signal processing on the received signals to visualize the shape, properties, or dynamics of the subject's interior as an ultrasound image. Ultrasound diagnostic devices can obtain ultrasound images through a simple operation of placing an ultrasound probe on the body surface or inserting it into the body, so they are safe and impose little strain on the subject.

[0003] Ultrasound diagnostic devices are used, for example, when treating a target area by inserting a puncture needle into the body of a subject under ultrasound guidance. In such treatment, a practitioner such as a doctor can insert the puncture needle and perform the treatment while checking the target area by looking at ultrasound images obtained by the ultrasound diagnostic device.

[0004] When performing treatment under ultrasound guidance, it is preferable that the target region be clearly reflected in the ultrasound image (B-mode image) in order to accurately grasp the position and area of ​​the treatment region. For example, in a nerve block, in which local anesthesia is administered by puncturing a peripheral nerve directly or in its vicinity, the target may be the nerve into which the anesthetic agent is to be injected or a blood vessel into which the anesthetic agent must not be accidentally injected. In addition, during a nerve block, the practitioner visually distinguishes between the nerve and the blood vessel on the ultrasound image and takes care not to puncture the blood vessel, but this requires high skill and extensive experience.

[0005] Against this background, in recent years, technologies have been proposed that identify targets within ultrasound images and provide a display image of the ultrasound image to a practitioner (hereinafter also referred to as a "user") in a manner that allows the area of ​​the target to be identified (see, for example, Patent Document 1 and Patent Document 2).

[0006] FIG. 1 is a diagram showing an example of an image processing method for an ultrasound image according to the prior art.

[0007] In a conventional image processing method, for example, a discrimination model trained by machine learning is used to identify a target (e.g., nervous tissue) in an ultrasound image, and a likelihood image is generated in which regions in the ultrasound image with a high likelihood (i.e., certainty) of the target being present are distinguished from regions with a low likelihood (i.e., certainty) (also referred to as segmentation processing).The conventional image processing method then uses a color map to add color information (hue, saturation, brightness) to each pixel of the ultrasound image based on the pixel values ​​of the ultrasound image and the pixel values ​​of the likelihood image, and changes at least one of the hue, saturation, and brightness of the ultrasound image to generate a display image to be provided to a user.

[0008] The "likelihood" of a target is an index showing the likelihood of it being a target, with the likelihood being high in the area where the target exists and low in the area where the target does not exist. The "likelihood image" is an image that shows the distribution of the likelihood of the target (i.e., the area where the target exists) corresponding to the entire ultrasound image. [Prior art documents] [Patent documents]

[0009] [Patent Document 1] Special Publication No. 2019-508072 [Patent Document 2] Patent Publication No. 2021-058232 Summary of the Invention [Problem to be solved by the invention]

[0010] Meanwhile, the inventors of the present application have been considering applying a spatial compounding method to this type of ultrasound diagnostic device in order to provide a user with high-quality ultrasound images and more accurately present the area of ​​a target (including not only a puncture target such as nervous tissue, but also an object such as a puncture needle that calls the user's attention to its presence area; the same applies hereinafter) in the ultrasound image. The spatial compounding method is a technique in which multiple frame images are generated by transmitting ultrasound beams from different directions toward the same location in a subject, and these multiple frame images are synthesized to generate a single spatial compound image.

[0011] FIG. 2 is a diagram illustrating a general spatial compounding method.

[0012] In the spatial compounding method, for example, as shown in Fig. 2, ultrasound image B generated by ultrasound scanning using an ultrasound beam with a steering angle of 0 degrees, ultrasound image A generated by ultrasound scanning using an ultrasound beam with a steering angle of -θ degrees, and ultrasound image C generated by ultrasound scanning using an ultrasound beam with a steering angle of +θ degrees are repeatedly generated in the same order in a three-frame cycle, and each time one frame of received data is acquired, ultrasound images for three frames are combined with the ultrasound images of the two immediately preceding frames to generate a spatial compound image Sy. In this way, the spatial compound image Sy, which is a combination of ultrasound images A, B, and C corresponding to three different steering angles, is constantly updated.

[0013] According to this spatial compounding method, by synthesizing multiple frame images generated by transmitting ultrasonic beams from different directions, it is possible to reduce speckle noise caused by scattered waves from countless scattering sources present within the subject, and also reduce acoustic noise such as shadows.

[0014] However, prior art spatial compounding methods have several problems that make the target area uncertain.

[0015] FIG. 3 is a diagram for explaining motion artifacts, which is one of the problems of the spatial compounding method according to the prior art.

[0016] Generally, in an ultrasound examination, a user moves an ultrasound probe along the surface of the subject's body to search for a treatment target (e.g., nervous tissue) present within the subject. As a result, the frame images of the target synthesized in each direction during spatial compounding processing are imaged at different positions. As a result, the spatial compound image generated by synthesizing the frame images in each direction becomes unclear, making it difficult to identify the target (nervous tissue HT in FIG. 3) from the spatial compound image. Note that such motion artifacts are generated not only by the movement of the ultrasound probe but also by the movement of the tissue itself (e.g., the heart) within the subject.

[0017] FIG. 4 is a diagram illustrating another problem of the spatial compounding method according to the prior art, that is, the increased difficulty in identifying structures that have acoustic reflection anisotropy (hereinafter simply referred to as "anisotropy") with respect to an ultrasonic beam.

[0018] In general, targets for which the user's attention is drawn in ultrasound examinations include both structures that are not anisotropic to the ultrasound beam, such as neural tissue (HT in Figure 4), and structures that are anisotropic to the ultrasound beam, such as the puncture needle (QT in Figure 4). Ultrasound generally reflects at boundaries where acoustic impedance differs, and the closer the angle to the boundary is to a 90-degree angle, the stronger the reflection, resulting in a clearer reflected ultrasound. Therefore, structures such as neural tissue that induce reflected ultrasound in various directions in response to an incident ultrasound beam do not depend on the beam direction of the ultrasound beam, and therefore there is no risk of the image being blurred on a spatial compound image. However, in the case of a puncture needle, when the beam direction of the ultrasound beam is perpendicular to the extension direction of the puncture needle, the puncture needle is clearly depicted in the ultrasound image. However, when the beam direction of the ultrasound beam is parallel to the extension direction of the puncture needle, the puncture needle is barely depicted in the ultrasound image.

[0019] In other words, in a method such as the spatial compounding method according to the prior art, in which a spatial compound image is generated by simply averaging frame images in each direction, the image synthesis results in blurred images of anisotropic structures such as a puncture needle, making it more difficult than usual to identify such structures in the spatial compound image.

[0020] FIG. 5 is a diagram illustrating another problem of the spatial compounding method according to the prior art, that is, the increased difficulty in identifying structures present at the edge of an image.

[0021] In general, in spatial compound processing, ultrasound images (ultrasound images A and C in Figure 2) generated by ultrasound scanning using ultrasound beams in the outer transmission directions are trimmed to fit the image area of ​​an ultrasound image generated by ultrasound scanning using an ultrasound beam with a steering angle of 0 degrees, and then these images are synthesized.

[0022] Generally, the more the target (nerve tissue HT in Figure 3) is visible in the ultrasound image around its entire periphery, the easier it is to identify the target, but if the target is located at the edge of the image and is depicted with a portion missing, the difficulty of identifying the target increases. In other words, what is identifiable in an ultrasound image generated by ultrasound scanning using ultrasound beams in the outer transmission directions (see ultrasound image A in Figure 5) becomes more difficult than usual in a spatial compound image generated by simply averaging frame images in each direction.

[0023] The present disclosure has been made in consideration of the above-mentioned problems, and aims to provide an ultrasound diagnostic device, a control method for an ultrasound diagnostic device, and a control program for an ultrasound diagnostic device that enable improvement in the accuracy of a likelihood image showing a target region in ultrasound image diagnosis using spatial compounding. [Means for solving the problem]

[0024] The present disclosure mainly solves the above-mentioned problems by: a transmitting / receiving unit that causes the ultrasonic probe to transmit and receive ultrasonic beams; a signal processing unit that generates an ultrasound image based on a received signal acquired from the ultrasound probe; a target identification unit that performs segmentation processing on the ultrasound image based on the structure type to generate a likelihood image that indicates a region where a target exists in the ultrasound image; a likelihood image synthesis unit that synthesizes the likelihood images of the respective ultrasound images generated by ultrasound scanning using the ultrasound beams with mutually different steering angles to generate a spatial compound likelihood image; The ultrasound diagnostic device is equipped with:

[0025] In other respects, A process of causing an ultrasonic probe to transmit and receive ultrasonic beams; A process of generating an ultrasound image based on a received signal acquired from the ultrasound probe; a process of performing a segmentation process on the ultrasound image based on a structure type to generate a likelihood image indicating an area where a target exists in the ultrasound image; a process of generating a spatial compound likelihood image by combining the likelihood images of the respective ultrasound images generated by ultrasound scanning using the ultrasound beams with mutually different steering angles; The present invention relates to a method for controlling an ultrasonic diagnostic apparatus having the above-mentioned features.

[0026] In other respects, On the computer, A process of causing an ultrasonic probe to transmit and receive ultrasonic beams; A process of generating an ultrasound image based on a received signal acquired from the ultrasound probe; a process of performing a segmentation process on the ultrasound image based on a structure type to generate a likelihood image indicating an area where a target exists in the ultrasound image; a process of generating a spatial compound likelihood image by combining the likelihood images of the respective ultrasound images generated by ultrasound scanning using the ultrasound beams with mutually different steering angles; This is a control program for an ultrasound diagnostic device that executes the above. [Effects of the Invention]

[0027] According to the ultrasound diagnostic device according to the present disclosure, it is possible to improve the accuracy of a likelihood image showing a target region in ultrasound image diagnosis using spatial compounding. [Brief explanation of the drawings]

[0028] [Figure 1] FIG. 1 is a diagram showing an example of an image processing method for an ultrasound image according to the prior art; [Figure 2] A diagram explaining the general spatial compounding method [Figure 3] 1 is a diagram illustrating motion artifacts, which are one of the problems of the spatial compounding method according to the prior art. [Figure 4] FIG. 1 is a diagram for explaining another problem of the spatial compounding method according to the prior art, that is, the increased difficulty in identifying structures that have acoustic reflection anisotropy with respect to ultrasonic beams. [Figure 5] 1 is a diagram illustrating another problem with the spatial compounding method according to the prior art, which is the increased difficulty of identifying structures present at the edge of an image. [Figure 6] FIG. 1 is a diagram showing an example of the appearance of an ultrasound diagnostic apparatus according to an embodiment of the present invention. [Figure 7] Block diagram showing the main parts of the control system of the ultrasound diagnostic device [Figure 8] A diagram showing the detailed configuration of an image processing unit. [Figure 9] FIG. 10 is a diagram illustrating the processing performed by the target identification unit. [Figure 10] FIG. 10 is a diagram illustrating the processing performed by the likelihood image synthesis unit. [Figure 11] FIG. 10 is a diagram showing an example of an image synthesis method according to an object to be identified stored in an image synthesis method data table. DETAILED DESCRIPTION OF THE INVENTION

[0029] Hereinafter, preferred embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. In this specification and drawings, components having substantially the same functions are designated by the same reference numerals, and redundant description will be omitted.

[0030] [Overall configuration of ultrasound diagnostic equipment] The overall configuration of an ultrasound diagnostic device according to one embodiment of the present invention (hereinafter referred to as "ultrasound diagnostic device 1") will be described below with reference to Figures 5 and 6. The ultrasound diagnostic device 1 according to this embodiment is used, for example, to visualize the shape, properties, or dynamics inside a subject as an ultrasound image and perform image diagnosis.

[0031] Fig. 6 is a diagram showing an example of the appearance of the ultrasonic diagnostic apparatus 1 according to this embodiment. Fig. 7 is a block diagram showing the main parts of the control system of the ultrasonic diagnostic apparatus 1 according to this embodiment.

[0032] The ultrasound diagnostic device 1 is used to visualize the shape, properties, or dynamics inside a subject as an ultrasound image and perform image diagnosis. For example, when performing a nerve block by puncturing the subject and injecting an anesthetic into a nerve or the area around the nerve, the ultrasound diagnostic device 1 has a function of visually displaying the area where a target exists so as to be superimposed on a B-mode image as puncture support information.

[0033] In this embodiment, for example, the nerve tissue for identifying the area where the puncture needle should be inserted, and the puncture needle itself, can be the "target" that alerts the user to the area where the puncture needle is present. However, the target setting can be changed arbitrarily depending on the user's usage of the ultrasound diagnostic device. Nerves may be treated as targets, and structures other than nerves, such as blood vessels, bones, and muscle fibers, may be treated as non-targets, or nerves and blood vessels into which the puncture needle should not be inserted may be treated as targets, and other structures may be treated as non-targets.

[0034] The ultrasonic diagnostic device 1 includes an ultrasonic diagnostic device main body 10 and an ultrasonic probe 20. The ultrasonic diagnostic device main body 10 and the ultrasonic probe 20 are connected via a cable 30, for example.

[0035] The ultrasonic probe 20 transmits ultrasonic waves to the subject, receives ultrasonic echoes reflected within the subject, converts them into received signals, and transmits them to the ultrasonic diagnostic apparatus main body 10. Any type of probe, such as a convex type, a linear type, or a sector type, can be used as the ultrasonic probe 20.

[0036] The ultrasonic probe 20 has an array transducer 21 composed of a plurality of piezoelectric transducers arranged in an array, and a channel switching unit (not shown) for individually switching on and off the driving state of each of the plurality of piezoelectric transducers that make up the array transducer 21.

[0037] The transducer array 21 is composed of, for example, a plurality of piezoelectric transducers arranged in an array along the scanning direction. The driving states of the plurality of piezoelectric transducers constituting the transducer array 21 are switched on and off individually or in blocks in order along the scanning direction under the control of the channel switching unit by the control unit 16. That is, the plurality of piezoelectric transducers individually or in blocks convert voltage pulses generated by the transmitter / receiver unit 11 into ultrasonic beams and transmit them into the subject, and also receive reflected wave beams generated when the ultrasonic beams are reflected within the subject, convert them into electrical signals, and output them to the transmitter / receiver unit 11. In this way, the ultrasonic probe 20 transmits and receives ultrasonic waves so as to scan the subject.

[0038] The ultrasound diagnostic device main body 10 includes a transmitting / receiving unit 11, a signal processing unit 12, an image processing unit 13, a display unit 14, an operation input unit 15, and a control unit 16.

[0039] The transmitting / receiving unit 11 is a transmitting / receiving circuit that causes the ultrasonic probe 20 to transmit and receive ultrasonic waves.

[0040] The transmitting / receiving unit 11 has a transmitting unit 11a that generates voltage pulses (hereinafter referred to as "drive signals") and sends them to the individual piezoelectric vibrators of the ultrasonic probe 20, and a receiving unit 11b that receives and processes electrical signals (hereinafter referred to as "received signals") related to reception beams generated by the individual piezoelectric vibrators of the ultrasonic probe 20. Then, under the control of the control unit 16, the transmitting unit 11a and the receiving unit 11b each perform an operation to cause the ultrasonic probe 20 to transmit and receive ultrasonic waves.

[0041] The transmitting unit 11a is configured to include, for example, a pulse oscillator and a pulse setting unit provided for each channel connected to the ultrasonic probe 20. The transmitting unit 11a adjusts the voltage pulse generated by the pulse oscillator to the voltage amplitude, pulse width, and timing set in the pulse setting unit, and sends it to the transducer array 21. The transmitting unit 11a sets an appropriate delay time for each channel so that the ultrasonic waves output from each piezoelectric transducer of the ultrasonic probe 20 are focused in a beam shape in a predetermined direction, and supplies a drive signal to each piezoelectric transducer.

[0042] The receiving unit 11b includes, for example, a preamplifier, an AD converter, and a receive beamformer. The preamplifier and the AD converter are provided for each channel connected to the ultrasonic probe 20, and amplify weak received signals and convert the amplified received signals (analog signals) into digital signals. The receive beamformer combines multiple received signals into one by phasing and adding the received signals (digital signals) of each channel, and outputs the combined signals to the signal processing unit 12. For example, the receive beamformer sets an appropriate delay time for each channel to focus ultrasonic echoes from a predetermined direction, and combines the multiple received signals into one and outputs the combined signals to the signal processing unit 12. In addition, the receive beamformer performs dynamic receive focus control so as to continuously move the receive focus point from the vicinity of the ultrasonic wave emitting surface of the ultrasonic probe 20 in a deeper direction.

[0043] The signal processing unit 12 detects (envelope detects) the sound ray data input from the receiving unit 11b to obtain a signal, and also performs logarithmic amplification, filtering (e.g., low-pass filtering, smoothing, etc.), emphasis processing, etc. as necessary. The signal processing unit 12 then sequentially accumulates the received signals at each scanning position in a frame memory, and generates two-dimensional data consisting of sampling data (e.g., signal intensity of received signals) at each position in a cross section along the scanning direction and depth direction. For example, the signal processing unit 12 converts the signal intensity of the received signals at each position in the two-dimensional data into pixel values, and generates one frame of ultrasound image data for B-mode display (hereinafter abbreviated as "ultrasound image"). The signal processing unit 12 then generates the ultrasound image each time the transmitting / receiving unit 11 scans the inside of the subject.

[0044] The signal processing unit 12 may have a quadrature detection processing unit, an autocorrelation calculation unit, etc. so that an ultrasonic image related to a Doppler image can be generated.

[0045] The image processing unit 13 performs spatial compounding processing on the ultrasound images generated by the signal processing unit 12, synthesizing multiple ultrasound images generated by ultrasound scanning using ultrasound beams with different steering angles to generate a single ultrasound image (hereinafter referred to as a "spatially compounded ultrasound image") for display.

[0046] Furthermore, the image processing unit 13 performs segmentation processing based on the structure type on the ultrasound images generated by the signal processing unit 12 to generate a likelihood image showing the area where the target exists. Then, the image processing unit 13 synthesizes the likelihood images of each of the multiple ultrasound images generated by ultrasound scanning using ultrasound beams with different steering angles to generate a single likelihood image (hereinafter referred to as a "spatial compound likelihood image") as an image to be displayed.

[0047] The transmitter / receiver 11, signal processor 12, and image processor 13 are configured with dedicated or general-purpose hardware (i.e., electronic circuits) for each process, such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field-Programmable Gate Array), and realize each function in cooperation with the controller 16. However, some or all of these may be realized by a DSP (Digital Signal Processor), a CPU (Central Processing Unit), a GPGPU (General-Purpose Graphics Processing Unit), or the like performing arithmetic processing according to a program.

[0048] The display unit 14 is, for example, a display such as an LCD (Liquid Crystal Display), etc. The display unit 14 acquires display image data from the image processing unit 13 and displays the display image data.

[0049] The operation input unit 15 is, for example, a keyboard or a mouse, and acquires an operation signal input by a user. The operation input unit 15 can set, for example, the type of ultrasound probe 20, the type of subject (i.e., the type of biological tissue), the depth of the imaging target within the subject, or the imaging mode (e.g., B mode, C mode, or E mode), based on the operation input by the user.

[0050] The control unit 16 controls the entire ultrasound diagnostic apparatus 1 by controlling the transmitting / receiving unit 11, the signal processing unit 12, the image processing unit 13, the display unit 14, and the operation input unit 15 according to their respective functions.

[0051] The control unit 16 includes, for example, a CPU (Central Processing Unit) as an arithmetic / control device, and a ROM (Read Only Memory) and RAM (Random Access Memory) as main storage devices. Basic programs and basic setting data are stored in the ROM. The CPU reads out a program corresponding to the processing content from the ROM, loads it into the RAM, and executes the loaded program, thereby centrally controlling the operation of each functional block of the ultrasound diagnostic apparatus main unit 10.

[0052] In this embodiment, the functions of each functional block are realized by cooperation between the hardware constituting the functional block and the control unit 16. However, the control unit 16 may execute a program to realize some or all of the functions of each functional block.

[0053] The control unit 16 determines the transmission and reception conditions of ultrasound in the ultrasound probe 20 (for example, aperture conditions, focal point, transmission waveform, center frequency or band, and apodization) based on the type of ultrasound probe 20 (for example, compex type, sector type, or linear type), the depth of the imaging target in the subject, and the imaging mode (for example, B mode, C mode, or E mode) set in the operation input unit 15. Then, the control unit 16 operates the transmission and reception unit 11 in accordance with the transmission and reception conditions of ultrasound in the ultrasound probe 20.

[0054] [Detailed configuration of image processing unit 13] FIG. 8 is a diagram showing a detailed configuration of the image processing unit 13 according to this embodiment.

[0055] The image processing unit 13 according to this embodiment includes a first DSC (Digital Scan Converter) 13a, an ultrasound image synthesis unit 13b, a target identification unit 13c, a second DSC (Digital Scan Converter) 13d, a likelihood image synthesis unit 13e, and a display image generation unit 13f.

[0056] The first DSC 13a performs coordinate conversion processing and pixel interpolation processing on the ultrasound image generated by the signal processing unit 12 according to the type of ultrasound probe 20, and converts the ultrasound image data into display image data that conforms to the scanning method of the television signal of the display unit 14.

[0057] As described above with reference to FIG. 2, the ultrasound image synthesis unit 13b synthesizes multiple ultrasound images generated by ultrasound scanning using ultrasound beams with mutually different steering angles to generate one spatial compound ultrasound image.

[0058] In the present embodiment, as an example, the steering angle of the ultrasonic beam transmitted from the ultrasonic probe 20 is controlled under the control of the control unit 16, and an ultrasonic image B generated by ultrasonic scanning using an ultrasonic beam with a steering angle of 0 degrees, an ultrasonic image A generated by ultrasonic scanning using an ultrasonic beam with a steering angle of -θ degrees, and an ultrasonic image C generated by ultrasonic scanning using an ultrasonic beam with a steering angle of +θ degrees are repeatedly generated in the same order in a three-frame cycle, as shown in Fig. 2. Every time one frame of ultrasonic image is acquired, the ultrasonic image synthesis unit 13b synthesizes three frames of ultrasonic images by combining the ultrasonic images of the two immediately preceding frames to generate a spatially compounded ultrasonic image Sy.

[0059] At this time, for example, the ultrasound image synthesis unit 13b trims the outer regions of ultrasound image A generated by ultrasound scanning using an ultrasound beam with a steering angle of -θ degrees and ultrasound image C generated by ultrasound scanning using an ultrasound beam with a steering angle of +θ degrees, which do not overlap with ultrasound image B generated by ultrasound scanning using an ultrasound beam with a steering angle of 0 degrees, and after unifying the coordinate systems of these ultrasound images A, B, and C, synthesizes these ultrasound images A, B, and C by a method of adding and averaging the portions where B-mode image signals acquired from the same position overlap.

[0060] The number of frames of the ultrasound images to be synthesized by the ultrasound image synthesis unit 13b may be other than three.

[0061] 1, the target identification unit 13c uses the identification model D1 to perform segmentation processing based on the structure type on the ultrasound image generated by the signal processing unit 12, and generates a likelihood image (an image showing the distribution of the likelihood of the target in the ultrasound image) showing the presence area of ​​the target in the ultrasound image. That is, the target identification unit 13c identifies the target (for example, nerve tissue or a puncture needle) in the ultrasound image.

[0062] Here, the discrimination model D1 is, for example, a neural network (e.g., a convolutional neural network) that has undergone a learning process in advance using a known machine learning algorithm (e.g., backpropagation) so as to extract features of an input ultrasound image and output a likelihood distribution of a target in the ultrasound image, and the learning process is pre-stored in a storage unit of the image processing unit 13. Such discrimination model D1 is typically constructed by supervised learning using training data configured from a data set in which ultrasound images are associated with likelihood distributions of targets. For an example of the learning process of the discrimination model D1, see, for example, Patent Document 2, a prior application of the applicant of the present application.

[0063] The identification model D1 has been trained to identify at least one structure type from an ultrasound image, for example, nerve tissue, vascular tissue, muscle tissue, fascial tissue, tendon tissue, or a puncture needle. A separate identification model D1 may be prepared for each structure type, or one identification model D1 may be configured to identify multiple structure types. Similarly, the target identification unit 13c may switch the type of identification model D1 depending on the type of target to be identified.

[0064] In other words, the identification model D1 calculates the likelihood of a target for each pixel or each pixel block (meaning a pixel group consisting of multiple pixels) in association with each pixel region in the ultrasound image, and outputs a distribution of target likelihoods (i.e., a likelihood image) corresponding to the entire input ultrasound image. The identification model D1 according to this embodiment is configured to output the likelihood of a target corresponding to a central pixel block of an input ultrasound image of a predetermined size. The target identification unit 13c then switches the input image for the identification model D1 so that it scans the entire ultrasound image by raster scanning for each predetermined size, thereby outputting a distribution of target likelihoods (i.e., a likelihood image) for the entire ultrasound image. At this time, the target identification unit 13c outputs the distribution of target likelihoods from the input image, for example, by forward propagation processing of the identification model D1 (neural network).

[0065] The likelihood image generated by the target identification unit 13c is, for example, data in which a likelihood of one value in the range of 0 to 1 is calculated for each pixel region of the ultrasound image (see FIG. 1). Such a likelihood image may, for example, show the distribution of the likelihood of one type of target (e.g., nervous tissue) in the entire ultrasound image, or may show the distribution of the likelihood of each of multiple types of targets (e.g., nervous tissue and a puncture needle) in the entire ultrasound image. Furthermore, the size of the likelihood image (i.e., the number of pixels) may be the same as the size of the ultrasound image, or may be scaled down compared to the size of the ultrasound image.

[0066] The discrimination model D1 used in the target discrimination unit 13c may be a discrimination model other than a neural network, such as a support vector machine (SVM), a k-nearest neighbor method, a random forest, or a combination of these. This type of discrimination model is subjected to a learning process to extract features of a pattern to be discriminated and is autonomously optimized so as to be able to accurately discriminate the pattern to be discriminated even from data superimposed with noise, etc., and is therefore useful in that it can constitute a highly robust discriminator.

[0067] The second DSC 13d performs coordinate conversion processing and pixel interpolation processing on the likelihood image generated by the target identification unit 13c according to the type of ultrasonic probe 20, and converts the likelihood image data into display image data that conforms to the scanning method of the television signal of the display unit 14.

[0068] The likelihood image synthesis unit 13e synthesizes likelihood images obtained from a plurality of ultrasound images generated by ultrasound scanning using ultrasound beams with mutually different steering angles, to generate a spatial compound likelihood image.

[0069] Basically, similar to the synthesis process of the ultrasound image synthesis unit 13b, the likelihood image synthesis unit 13e unifies the coordinate systems of the likelihood images and then synthesizes the likelihood images by averaging the likelihoods at the same position of the likelihood images, etc. However, the likelihood image synthesis unit 13e refers to an image synthesis method data table D2 stored in advance in a storage unit (not shown) of the image processing unit 13, and synthesizes the likelihood images using an image synthesis method set for each structure type (see FIG. 11).

[0070] Fig. 9 is a diagram illustrating the processing performed by the target identification unit 13c according to this embodiment. Fig. 10 is a diagram illustrating the processing performed by the likelihood image synthesis unit 13e according to this embodiment. Fig. 11 is a diagram illustrating an example of an image synthesis method according to the identification target stored in the image synthesis method data table D2.

[0071] 9, the target identifier 13c according to this embodiment performs segmentation processing based on the structure type on each of the ultrasound images sequentially generated by the signal processor 12 to generate likelihood images indicating the presence area of ​​the target. That is, the target identifier 13c performs the identification processing on ultrasound image B generated by ultrasound scanning using an ultrasound beam with a steering angle of 0 degrees to generate likelihood image B1, performs the identification processing on ultrasound image A generated by ultrasound scanning using an ultrasound beam with a steering angle of -θ degrees to generate likelihood image A1, and performs the identification processing on ultrasound image C generated by ultrasound scanning using an ultrasound beam with a steering angle of +θ degrees to generate likelihood image C1. Here, the identification model D1 applied to ultrasound image A, ultrasound image B, and ultrasound image C may be the same or different and optimized for each steering angle.

[0072] This processing by the target identification unit 13c is not affected by motion artifacts caused by spatial compound synthesis, so the target identification unit 13c can generate likelihood images A1, B1, and C1 from ultrasound image A, B, and C, respectively, in which the likelihood of the target region is calculated with high accuracy.

[0073] Furthermore, the processing by the target identification unit 13c allows for highly accurate identification of structures that have acoustic reflection anisotropy with respect to the ultrasound beam, such as the puncture needle QT, in any of the likelihood images A1, B1, or C1. In Figure 9, when the ultrasound beam is steered at a +θ angle, the direction is perpendicular to the extension direction of the puncture needle QT, and the puncture needle QT is clearly depicted in the ultrasound image C and the likelihood image C1.

[0074] Furthermore, the processing by target identification unit 13c allows structures present at the edge of the image to be identified with high accuracy in likelihood image A1, likelihood image B1, or likelihood image C1. For example, in Figure 5, in ultrasound image A, nervous tissue HT is clearly depicted, and likelihood image A1 can be generated in which the likelihood of the target region is calculated with high accuracy.

[0075] Then, the likelihood image synthesis unit 13e in this embodiment synthesizes multiple likelihood images A1, B1, and C1 to generate a spatial compound likelihood image Sy1 using an image synthesis method set for each structure type stored in the image synthesis method data table D2, as shown in Figure 11.

[0076] For example, when the target to be identified is a structure having acoustic reflection anisotropy with respect to the ultrasonic beam (e.g., a puncture needle QT), the likelihood image composition unit 13e selects, for each pixel region, the maximum likelihood among the likelihoods of the plurality of likelihood images A1, B1, and C1 to be combined, or selectively adds, for each pixel region, the likelihoods of the plurality of likelihood images A1, B1, and C1 to be combined that are equal to or greater than a threshold. Furthermore, when the target to be identified is a structure having no acoustic reflection anisotropy with respect to the ultrasonic beam (e.g., nervous tissue HT), the likelihood image composition unit 13e averages, for each pixel region, the likelihoods of the plurality of likelihood images A1, B1, and C1 to be combined, to combine the plurality of likelihood images A1, B1, and C1.

[0077] Examples of structures that have acoustic reflection anisotropy with respect to ultrasonic beams include puncture needles and fascia, while examples of structures that do not have acoustic reflection anisotropy with respect to ultrasonic beams include nerve tissue and muscle tissue.

[0078] When there is one type of target, likelihood image composition unit 13e may compose likelihood images A1, B1, and C1 for the entire region of the likelihood image using one type of image composition method corresponding to the structure type of the target. On the other hand, when there are multiple types of targets (i.e., when generating likelihood images showing the distribution of likelihoods of multiple types of targets), likelihood image composition unit 13e may compose multiple likelihood images for each pixel region of the likelihood image using an image composition method corresponding to the type of target present in that pixel region.

[0079] 9 shows an example in which two types of targets, the nervous tissue HT and the puncture needle QT, are set as targets to be identified. Each pixel value in the nervous tissue HT region of the spatial compound likelihood image Sy1 is calculated as the average value of the likelihoods of the likelihood images A1, B1, and C1, and each pixel value in the puncture needle QT region of the spatial compound likelihood image Sy1 is calculated by selecting the maximum value of the likelihoods of the likelihood images A1, B1, and C1.

[0080] According to the processing of the likelihood image composition unit 13e, likelihood image A1, likelihood image B1, and likelihood image C1, for which the likelihood of the target region has been calculated with high accuracy, can be combined to generate a spatial compound likelihood image Sy1, so that the likelihood image composition unit 13e can construct a highly accurate likelihood image (i.e., likelihood distribution) in which the influence of motion artifacts is suppressed. This is because it is possible to avoid the difficulty of identifying the target using a discriminative model from a spatial compound ultrasound image that has been generated with blurring due to the influence of motion artifacts.

[0081] Furthermore, according to the processing of this likelihood image synthesis unit 13e, multiple likelihood images A1, B1, and C1 are synthesized using an image synthesis method set for each structure type to generate a spatial compound likelihood image Sy1, so that, for example, information on a likelihood image from likelihood image A1, likelihood image B1, or likelihood image C1 that clearly depicts a structure (here, the puncture needle QT) that has acoustic reflection anisotropy for an ultrasound beam appears clearly on the spatial compound likelihood image Sy1. In other words, this makes it possible to clarify the presence area of ​​the target on the spatial compound likelihood image Sy1.

[0082] Furthermore, according to the processing of the likelihood image composition unit 13e, likelihood images A1, B1, and C1 are generated from the ultrasound images A, B, and C, respectively, and then these are composed, so that targets present at the image edges can also be identified with high accuracy. This is because at least one of ultrasound image A, ultrasound image B, and ultrasound image C shows the entire structure present at the image edge, and at least one of likelihood image A1, likelihood image B1, and likelihood image C1 generated from such ultrasound image A, ultrasound image B, and ultrasound image C can identify targets with high accuracy.

[0083] It is preferable that the likelihood image synthesis unit 13e removes noise contained in each of the likelihood images A1, B1, and C1 based on changes in information related to likelihood (e.g., target likelihood, likelihood distribution) obtained from temporally consecutive ultrasound images before synthesizing the likelihood images A1, B1, and C1. In this case, it is preferable that the likelihood image synthesis unit 13e separates, for example, the temporal change in the likelihood image B1 obtained from the ultrasound image B generated by ultrasound scanning using an ultrasound beam with a steering angle of 0 degrees, the temporal change in the likelihood image A1 obtained from the ultrasound image A generated by ultrasound scanning using an ultrasound beam with a steering angle of −θ degrees, and the temporal change in the likelihood image C1 obtained from the ultrasound image C generated by ultrasound scanning using an ultrasound beam with a steering angle of +θ degrees, and performs noise processing.

[0084] In this case, the likelihood image synthesis unit 13e can remove noise contained in the likelihood image by, for example, applying moving average filtering or median filtering in the time axis direction. In this case, a region where the change (steepness) of the likelihood information exceeds a preset threshold may be detected as a noise region, and noise removal processing may be performed only on this noise region.

[0085] Furthermore, when likelihood image synthesis unit 13e synthesizes likelihood image A1, likelihood image B1, and likelihood image C1, likelihood image synthesis unit 13e may perform normalization processing on likelihood image A1, likelihood image B1, and likelihood image C1 as preprocessing.

[0086] The display image generating unit 13f applies the spatial compound likelihood image Sy1 generated by the target identifying unit 13c as an enhancement map of the target region in the spatial compound ultrasound image Sy generated by the ultrasound image combining unit 13b.

[0087] The display image generating unit 13f, for example, superimposes the spatial compound likelihood image Sy1 generated in the target identifying unit 13c on the spatial compound ultrasound image Sy generated in the ultrasound image combining unit 13b and outputs it to the display unit 14.

[0088] At this time, the display image generation unit 13f may combine the spatial compound ultrasound image Sy and the spatial compound likelihood image Sy1 using, for example, the color map shown in Fig. 1. The display image generation unit 13f adds color information (hue, saturation, brightness) to each pixel of the spatial compound ultrasound image Sy based on, for example, the pixel values ​​of the spatial compound ultrasound image Sy and the pixel values ​​of the spatial compound likelihood image Sy1 that are in a corresponding positional relationship within the image, and changes at least one of the hue, saturation, and brightness of the spatial compound ultrasound image Sy to generate a display image to be provided to the user.

[0089] In addition, instead of superimposing the spatial compound likelihood image Sy1 generated in the target identification unit 13c on the spatial compound ultrasound image Sy generated in the ultrasound image synthesis unit 13b, the display image generation unit 13f may display the spatial compound ultrasound image Sy and the spatial compound likelihood image Sy1 side by side.

[0090] [effect] As described above, the ultrasound diagnostic device 1 according to this embodiment: a transmitting / receiving unit 11 that causes the ultrasonic probe 20 to transmit and receive ultrasonic beams; a signal processing unit 12 that generates an ultrasonic image based on a received signal acquired from the ultrasonic probe 20; a target identification unit (13c) that performs segmentation processing on the ultrasonic image based on the type of structure to generate a likelihood image that indicates an area where a target exists in the ultrasonic image; a likelihood image synthesis unit 13e that synthesizes the likelihood images of the plurality of ultrasound images generated by ultrasound scanning using ultrasound beams with mutually different steering angles to generate a spatial compound likelihood image; It is equipped with:

[0091] Therefore, the ultrasound diagnostic device 1 according to this embodiment can construct a highly accurate likelihood image (i.e., likelihood distribution) in which the influence of motion artifacts is suppressed. This also makes it possible to identify targets that appear at the edge of the image with high accuracy. In other words, this makes it possible to more clearly identify the area in which the target exists on the likelihood image (here, the spatial compound likelihood image Sy1).

[0092] Furthermore, in the ultrasound diagnostic device 1 according to this embodiment, the likelihood image synthesis unit 13e in particular synthesizes multiple likelihood images to generate a spatial compound likelihood image using an image synthesis method set for each structure type.

[0093] Therefore, the ultrasound diagnostic device 1 according to this embodiment can accurately identify structures (such as a puncture needle or fascia) that have acoustic reflection anisotropy for ultrasound beams, thereby making it possible to more clearly identify the target's presence region on the likelihood image (here, the spatial compound likelihood image Sy1).

[0094] (Variation) In the above embodiment, the likelihood image synthesis unit 13e is configured to synthesize multiple likelihood images A1, B1, and C1 to generate a spatial compound likelihood image Sy1 using an image synthesis method set for each structure type pre-stored in the image synthesis method data table D2 (see Figure 11).

[0095] However, the image synthesis method in likelihood image synthesis unit 13e may be set by the user, which allows for more flexible processing.

[0096] The likelihood image composition unit 13e may allow the user to set the image composition method for each individual target appearing in the ultrasound image or for each pixel region of the ultrasound image. This allows, for example, for a target appearing at the edge of the image, the maximum value of the likelihoods of the likelihood images A1, B1, and C1 to be selected, while for a target appearing in the center of the image, the average value of the likelihoods of the likelihood images A1, B1, and C1 to be calculated. In this case, a user interface image may be displayed on the display unit 14 so that the user can selectively set the image composition method for a specific pixel region in the ultrasound image.

[0097] Although specific examples of the present invention have been described above in detail, these are merely examples and do not limit the scope of the claims. The technology described in the claims includes various modifications and alterations of the specific examples exemplified above. [Industrial Applicability]

[0098] According to the ultrasound diagnostic device according to the present disclosure, it is possible to improve the accuracy of a likelihood image showing a target region in ultrasound image diagnosis using spatial compounding. [Explanation of symbols]

[0099] 1. Ultrasound diagnostic equipment 10. Ultrasound diagnostic device body 11 Transmitter / Receiver 11a Transmitter 11b Receiving section 12 Signal Processing Section 13 Image processing section 13a 1st DSC 13b Ultrasound image synthesis unit 13c Target Identification Unit 13d 2nd DSC 13e Likelihood image synthesis unit 13f Display image generation section 14 Display section 15 Operation input section 16 Control Unit 20 Ultrasound Probe 30 Cable

Claims

1. a transmitting / receiving unit that causes the ultrasonic probe to transmit and receive ultrasonic beams; a signal processing unit that generates an ultrasound image based on a received signal acquired from the ultrasound probe; a target identification unit that performs segmentation processing on the ultrasound image based on the structure type to generate a likelihood image that indicates a region where a target exists in the ultrasound image; a likelihood image synthesis unit that synthesizes the likelihood images of the plurality of ultrasound images generated by ultrasound scanning using the ultrasound beams with mutually different steering angles to generate a spatial compound likelihood image; An ultrasound diagnostic device comprising: the target includes at least a first structure having acoustic reflection anisotropy with respect to the ultrasonic beam and a second structure having no acoustic reflection anisotropy with respect to the ultrasonic beam; The likelihood image synthesis unit is an ultrasound diagnostic device that synthesizes multiple likelihood images using different image synthesis methods set for each type of target for each pixel region of the likelihood image to generate the spatial compound likelihood image.

2. The spatial compound likelihood image is applied as an enhancement map of the target region in a spatial compound ultrasound image generated by combining a plurality of the ultrasound images generated by ultrasound scanning using the ultrasound beams at mutually different steering angles. The ultrasonic diagnostic apparatus according to claim 1 .

3. the likelihood image synthesis unit synthesizes the plurality of likelihood images using an image synthesis method set for each of the structure types to generate the spatial compound likelihood image; The ultrasonic diagnostic apparatus according to claim 1 .

4. When multiple types of targets are set, the likelihood image synthesis unit synthesizes the plurality of likelihood images for each pixel region of the likelihood image using an image synthesis method according to the type of the target present in the pixel region, to generate the spatial compound likelihood image; The ultrasonic diagnostic apparatus according to claim 1 .

5. The likelihood image synthesis unit When the target to be identified is the first structure having acoustic reflection anisotropy with respect to the ultrasonic beam, the likelihood images are synthesized by selecting, for each pixel region, a likelihood that is maximum among the likelihoods of the plurality of likelihood images to be synthesized, or by selectively adding likelihoods that are equal to or greater than a threshold among the likelihoods of the plurality of likelihood images to be synthesized, When the target to be identified is the second structure that does not have acoustic reflection anisotropy with respect to the ultrasonic beam, the likelihood of each of the plurality of likelihood images to be synthesized is averaged for each pixel region, thereby synthesizing the plurality of likelihood images. The ultrasonic diagnostic apparatus according to claim 1 .

6. the first structure includes a puncture needle; the second structure comprises neural tissue. The ultrasonic diagnostic apparatus according to claim 1 .

7. the likelihood image synthesis unit synthesizes the plurality of likelihood images using an image synthesis method set by a user to generate the spatial compound likelihood image; The ultrasonic diagnostic apparatus according to claim 1 .

8. the target identification unit performs segmentation processing based on the structure type for each of the plurality of ultrasound images using an identification model trained by machine learning; The ultrasonic diagnostic apparatus according to claim 1 .

9. the discriminative model is a neural network; The ultrasonic diagnostic apparatus according to claim 8.

10. a first process for causing an ultrasonic probe to transmit and receive an ultrasonic beam; a second process for generating an ultrasound image based on the received signal acquired from the ultrasound probe; a third process of performing a segmentation process on the ultrasound image based on a structure type to generate a likelihood image indicating a region where a target exists in the ultrasound image; a fourth process of generating a spatial compound likelihood image by combining the likelihood images of the respective ultrasound images generated by ultrasound scanning using the ultrasound beams with different steering angles; A method for controlling an ultrasound diagnostic apparatus comprising: the target includes at least a first structure having acoustic reflection anisotropy with respect to the ultrasonic beam and a second structure having no acoustic reflection anisotropy with respect to the ultrasonic beam; In the fourth processing, a control method is used in which multiple likelihood images are synthesized to generate the spatial compound likelihood image using different image synthesis methods set for each type of target for each pixel region of the likelihood image.

11. On the computer, a first process for causing an ultrasonic probe to transmit and receive an ultrasonic beam; a second process for generating an ultrasound image based on the received signal acquired from the ultrasound probe; a third process of performing a segmentation process on the ultrasound image based on a structure type to generate a likelihood image indicating a region where a target exists in the ultrasound image; a fourth process of generating a spatial compound likelihood image by combining the likelihood images of the respective ultrasound images generated by ultrasound scanning using the ultrasound beams with different steering angles; A control program for an ultrasonic diagnostic apparatus that executes the target includes at least a first structure having acoustic reflection anisotropy with respect to the ultrasonic beam and a second structure having no acoustic reflection anisotropy with respect to the ultrasonic beam; In the fourth process, a control program is provided that synthesizes multiple likelihood images to generate the spatial compound likelihood image for each pixel region of the likelihood image using a different image synthesis method set for each type of target.

Images