Ultrasonic diagnostic device

Abstract

To reduce the number of times of transmitting ultrasonic pulses in Doppler measurement while suppressing unnecessary components included in the information acquired by the Doppler measurement.SOLUTION: An information processing unit 26 includes a Doppler measurement part 40 for generating Doppler measurement information on the basis of a reception Doppler signal output from a wall filter 38, and a machine learning model 44 constructed on the basis of training information and target information. The training information is the information acquired from K reception Doppler signals. The target information is the information acquired from J reception Doppler signals on the same measurement object as the measurement object acquired from the K reception Doppler signals. J and K are integers of 2 or more, and J is larger than K. The machine learning model 44 generates Doppler measurement information on the basis of input information. The input information is the information generated from the K reception Doppler signals for the measurement object in a subject 18.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an ultrasonic diagnostic apparatus, and more particularly to an apparatus for performing Doppler measurement.

Background Art

[0002] An ultrasonic diagnostic apparatus that observes a subject by transmitting and receiving ultrasonic waves is widely used. Among the operation modes of the ultrasonic diagnostic apparatus, there is a Doppler mode for measuring the blood flow velocity at the observation site. In the measurement in the Doppler mode, ultrasonic pulses are transmitted to the subject over a plurality of times, and reflected waves generated in the subject over a plurality of times are received. A plurality of received signals having Doppler frequency components (hereinafter referred to as received Doppler signals) are generated from a plurality of received pulse signals based on the reflected waves received over a plurality of times. Further, high-pass filter processing (wall filter processing) is performed on the plurality of received Doppler signals, and based on the plurality of received Doppler signals after the high-pass filter processing, the blood flow velocity in the region of interest set on the ultrasonic beam formed by the ultrasonic pulses is obtained.

[0003] As the high-pass filter, a filter such as an FIR filter (Finite Impulse Response Filter) or an IIR filter (Infinite Impulse Response Filter) is used. The larger the number of received Doppler signals to be subjected to high-pass filter processing, the larger the number of tap coefficients of the high-pass filter can be increased. Therefore, the larger the number of received Doppler signals to be subjected to high-pass filter processing, the steeper the filter characteristics, and the larger the attenuation amount in the low frequency band can be made while securing the high frequency pass band. As a result, noise such as clutter in the displayed ultrasonic image is suppressed.

[0004] Note that, as technologies related to the present invention, Patent Documents 1 to 3 below describe ultrasonic diagnostic apparatuses operating in the Doppler mode. Patent Documents 4 to 6 have descriptions related to wall filters.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Patent Document 2

Patent Document 3

Patent Document 4

Patent Document 5

Patent Document 6

Summary of the Invention

Problems to be Solved by the Invention

[0006] As described above, in Doppler measurement, the more times an ultrasonic pulse is transmitted, the more received Doppler signals there are, the steeper the filter characteristics become, and the effect of suppressing unnecessary components such as clutter can be increased. However, the more times an ultrasonic pulse is transmitted, the lower the frame rate becomes, and the fewer the number of ultrasonic images displayed per unit time.

[0007] An object of the present invention is to reduce the number of times an ultrasonic pulse is transmitted in Doppler measurement while suppressing unnecessary components included in the information obtained by Doppler measurement.

Means for Solving the Problems

[0008] The ultrasonic diagnostic apparatus according to the present invention includes a transmission unit that transmits ultrasonic pulses N times by an ultrasonic probe, where N is an integer of 2 or more, a reception unit that receives reflected waves generated N times in a measurement object by the ultrasonic probe, and an information processing unit that generates N reception Doppler signals from N reception pulse signals output from the reception unit in response to the reflected waves generated N times in the measurement object and executes processing for each of the reception Doppler signals. The information processing unit includes a filter that performs high-pass filter processing on the N reception Doppler signals, a Doppler measurement unit that generates Doppler measurement information of a measurement object based on the N reception Doppler signals subjected to the high-pass filter processing, and a machine learning model constructed based on teacher data. Under the condition that J and K are integers of 2 or more and J is greater than K, the teacher data includes reception characteristic information derived from each of the reception Doppler signals with N being K, and training information that is at least one of the Doppler measurement information generated by the Doppler measurement unit when N is K, and reception characteristic information derived from each of the reception Doppler signals with N being J and target information that is at least one of the Doppler measurement information generated by the Doppler measurement unit when N is J for the same measurement object as when N is K. The machine learning model generates the Doppler measurement information based on the reception characteristic information derived from each of the reception Doppler signals with N being K and input information that is at least one of the Doppler measurement information generated by the Doppler measurement unit when N is K for a measurement object in a subject.

[0009] In one embodiment, the reception characteristic information includes at least one of the N reception Doppler signals before the high-pass filter processing and the N reception Doppler signals after the high-pass filter processing.

[0010] In one embodiment, the Doppler measurement information includes at least one of the autocorrelation values of the N received Doppler signals after the high-pass filter processing, the velocity of the measurement object obtained from the N received Doppler signals after the high-pass filter processing, the Doppler frequency variation degree of the N received Doppler signals after the high-pass filter processing, and the value indicating the magnitude of the N received Doppler signals after the high-pass filter processing.

[0011] In one embodiment, the transmitting unit transmits ultrasonic waves for B-mode image generation by the ultrasonic probe, the receiving unit receives the reflected waves for B-mode image generation generated by the measurement object by the ultrasonic probe, and the information processing unit includes a B-mode image generation unit that generates B-mode image data based on the B-mode image reception signal output from the receiving unit in response to the reflected waves for B-mode image generation. The B-mode image reception signal output from the receiving unit is used as any one of the N received pulse signals output from the receiving unit in response to the reflected waves for B-mode image generation.

[0012] In one embodiment, for the ultrasonic pulse, the transmitting unit forms a plurality of transmission bands at different positions or in different directions, and for two adjacent transmission bands, the ultrasonic pulse is transmitted alternately such that the ultrasonic pulse is transmitted in one transmission band while not being transmitted in the other transmission band. For two adjacent transmission band pairs each including two adjacent transmission bands, a series of the ultrasonic pulses transmitted alternately in the two transmission bands included in one transmission band pair is defined as an ultrasonic pulse group, and the ultrasonic pulse groups are transmitted alternately such that the ultrasonic pulse group is transmitted in one transmission band pair while not being transmitted in the other transmission band pair. Each time K ultrasonic pulses are transmitted in each transmission band, the machine learning model generates the Doppler measurement information for one frame for each transmission band.

[0013] In one embodiment, while the Doppler measurement information for one frame is being generated, for two adjacent transmission band pairs, an operation of alternately transmitting ultrasonic pulse groups is performed multiple times.

Advantages of the Invention

[0014] According to the present invention, it is possible to reduce the number of times of transmitting ultrasonic pulses in Doppler measurement while suppressing unnecessary components included in the information obtained by Doppler measurement.

Brief Description of the Drawings

[0015]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Embodiments for Carrying Out the Invention

[0016] Each embodiment of the present invention will be described with reference to each drawing. The same components shown in a plurality of drawings are denoted by the same reference numerals and their description will be omitted. FIG. 1 shows the configuration of an ultrasonic diagnostic apparatus 100 according to an embodiment of the present invention. The ultrasonic diagnostic apparatus 100 includes a beam control unit 10, a transmission unit 12, an ultrasonic probe 14, a reception unit 20, a control unit 22, an operation unit 24, an information processing unit 26, and a display unit 46.

[0017] The information processing unit 26 includes an integral addition unit 28, a B-mode image generation unit 32, an image processing unit 34, a quadrature detection unit 36, a wall filter 38, a Doppler measurement unit 40, and a machine learning unit 42. The information processing unit 26 may be composed of one or more processors that realize the respective functions of these components (the integral addition unit 28, the B-mode image generation unit 32, the image processing unit 34, the quadrature detection unit 36, the wall filter 38, the Doppler measurement unit 40, and the machine learning unit 42) by executing a program. The control unit 22 performs overall control of the ultrasonic diagnostic apparatus 100. The operation unit 24 includes a keyboard, a mouse, a lever, buttons, a voice recognition device, etc., and outputs information regarding the user's operation to the control unit 22. The control unit 22 controls the ultrasonic diagnostic apparatus 100 according to the operation in the operation unit 24.

[0018] The operation unit 24 may constitute a user interface for operating the ultrasonic diagnostic apparatus 100 together with the display unit 46. For example, the operation unit 24 may include a touch panel on the display screen of the display unit 46. Further, the operation unit 24 may have a function of operating a virtual keyboard, lever, buttons, etc. configured on the display screen.

[0019] An outline of the operation of the ultrasonic diagnostic apparatus 100 will be described. The ultrasonic diagnostic apparatus 100 transmits ultrasonic waves from the ultrasonic probe 14 to the subject 18, and receives the ultrasonic waves reflected within the subject 18, that is, the reflected waves generated within the subject 18, by the ultrasonic probe 14. The ultrasonic diagnostic apparatus 100 executes each measurement in the B-mode and the Doppler mode.

[0020] In the B-mode operation, the ultrasonic diagnostic apparatus 100 generates B-mode image data based on the reflected waves received from the subject 18, and displays the B-mode image on the display unit 46. In the Doppler-mode operation, the ultrasonic diagnostic apparatus 100 generates B-mode color Doppler image data indicating an image in which colors corresponding to the blood flow velocity are attached to the B-mode image, and displays the B-mode color Doppler image on the display unit 46.

[0021] The specific configuration of the ultrasonic diagnostic apparatus 100 and the specific processes executed by the ultrasonic diagnostic apparatus 100 will be described. The ultrasonic probe 14 is in a state of being in contact with the surface of the subject 18. The ultrasonic probe 14 includes a plurality of vibration elements 16. The transmission unit 12 outputs a transmission signal to each vibration element 16 of the ultrasonic probe 14 based on the control by the beam control unit 10. As a result, ultrasonic waves are transmitted from the ultrasonic probe 14. The beam control unit 10 may form a transmission beam in the ultrasonic probe 14 and scan the transmission beam with respect to the subject 18 by controlling the transmission unit 12. That is, the transmission unit 12 may adjust the delay time or level of each transmission signal according to the control of the beam control unit 10, form a transmission beam in the ultrasonic probe 14, and scan the transmission beam with respect to the subject 18. Further, the transmission unit 12 may adjust the delay time or level of each transmission signal according to the control of the beam control unit 10 and form a transmission band composed of a plurality of transmission beams in the ultrasonic probe 14.

[0022] When the reflected waves generated within the subject 18 are received by each vibration element 16 of the ultrasonic probe 14, each vibration element 16 outputs a reception signal corresponding to the received ultrasonic wave to the reception unit 20. The reception unit 20 performs processes such as amplification and frequency band limitation on the reception signals output from each vibration element 16 according to the control by the beam control unit 10 and outputs them to the information processing unit 26. The information processing unit 26 performs the processes executed by each component (the phased addition unit 28, the B-mode image generation unit 32, the image processing unit 34, the quadrature detection unit 36, the wall filter 38, the Doppler measurement unit 40, and the machine learning unit 42) on each reception signal.

[0023] The phased addition unit 28 performs phased addition on the plurality of received signals output from the receiving unit 20 for the plurality of vibration elements 16 to generate a phased addition received signal. As a result, a phased addition received signal is generated in which the phases are adjusted and added so that the received signals based on the ultrasonic waves received from a specific direction reinforce each other, and a reception beam is formed in that specific direction. The phased addition unit 28 outputs the phased addition received signal to the B-mode image generation unit 32 during operation in the B mode. Also, the phased addition unit 28 outputs the phased addition received signal to the quadrature detection unit 36 during operation in the Doppler mode.

[0024] The operation in the B mode will be described. The phased addition unit 28 generates each phased addition received signal based on the ultrasonic waves received from each direction of the reception beam scanned within the subject 18, and outputs it to the B-mode image generation unit 32 as a B-mode image reception signal. The B-mode image generation unit 32 generates B-mode image data based on the B-mode image reception signals obtained for each reception beam direction, and outputs it to the image processing unit 34. The B-mode image data based on one scan of the transmission beam and the reception beam (hereinafter referred to as the transmission / reception beam) is image data for one frame and corresponds to one B-mode image.

[0025] The beam control unit 10, the transmission unit 12, the ultrasonic probe 14, the receiving unit 20, the phased addition unit 28, and the B-mode image generation unit 32 successively generate B-mode image data along with the repeated scanning of the transmission / reception beam, and output each B-mode image data to the image processing unit 34. The image processing unit 34 generates a video signal for displaying a B-mode image based on the B-mode image data, and outputs it to the display unit 46. The display unit 46 displays a B-mode image based on the video signal.

[0026] The B mode may be the phase inversion B mode. In the phase inversion B mode, ultrasonic pulses are transmitted in the same direction twice with the phase shifted by 180°. The B mode image generation unit 32 adds two coherent received signals based on the reflected waves received twice according to the ultrasonic pulses transmitted twice to generate a harmonic reception signal, and generates phase inversion B mode image data based on the harmonic reception signal. The harmonic reception signal indicates the harmonic components generated in the two coherent received signals due to the Doppler frequency not being zero. The phase inversion B mode image data indicates an image of a moving measurement object such as blood.

[0027] The operation in the Doppler mode will be described. The operation of the Doppler mode according to this embodiment includes a basic operation, a learning operation, and a model application operation. In the basic operation, the Doppler measurement unit 40 generates Doppler measurement information based on the received Doppler signal obtained by quadrature detecting the coherent received signal.

[0028] The Doppler measurement information may include color Doppler data. The color Doppler data is data that indicates the blood flow velocity by color on the B mode image of the area scanned by the transmit-receive beam. The color Doppler data may be, for example, data that assigns blue to the area where blood flows away from the ultrasonic probe 14, assigns red to the area where blood flows towards the ultrasonic probe 14, and further increases the brightness in the area where the blood flow velocity is greater.

[0029] In addition to the color Doppler data, the Doppler measurement information generated by the Doppler measurement unit 40 may include various information related to the blood flow velocity in the area scanned by the transmit-receive beam. Information other than the color Doppler data included in the Doppler measurement information will be described later.

[0030] The learning operation is an operation of generating teacher data for machine learning and constructing a machine learning model 44 in the machine learning unit 42 by machine learning. The teacher data is generated by the basic operation. In the model application operation, Doppler measurement information is generated by the machine learning model 44 constructed by the machine learning unit 42 through machine learning.

[0031] The basic operation will be described. In the basic operation, ultrasonic pulses are transmitted and received N times for one transmission / reception beam. Here, N is an integer of 2 or more. From the coherent addition unit 28 to the quadrature detection unit 36, N received pulse signals are output as N coherent received signals. The quadrature detection unit 36 quadrature-detects the N received pulse signals with a local signal having the frequency of the transmitted ultrasonic pulse to generate N received Doppler signals. The N received Doppler signals are output from the quadrature detection unit 36 to the wall filter 38. The wall filter 38 performs high-pass filtering on the N received Doppler signals and outputs the N received Doppler signals after the high-pass filtering to the Doppler measurement unit 40.

[0032] Based on the N received Doppler signals acquired for each of a plurality of transmission / reception beams with different positions or directions, the Doppler measurement unit 40 generates color Doppler data as one piece of information included in the Doppler measurement information.

[0033] The Doppler measurement unit 40 obtains the blood flow velocity v(r) in the depth direction of the transmission / reception beam according to, for example, the autocorrelation processing shown in Patent Document 3 based on the N received Doppler signals. Here, v(r) indicates the blood flow velocity at the position of depth r. The Doppler measurement unit 40 generates color Doppler data based on the blood flow velocity v(r) in the depth direction at each depth r obtained for each of a plurality of transmission / reception beams with different positions or directions and outputs it to the image processing unit 34. Hereinafter, the specific processing for obtaining the blood flow velocity v(r) will be described.

[0034] The received Doppler signal acquired as the q-th (q = 1 to N) from the first to the N-th is represented by the following (Equation 1).

[0035]

Number

[0036] r is the coordinate value in the depth direction. I(r, q) represents the in-phase component of the received Doppler signal, and Q(r, q) represents the quadrature component of the received Doppler signal. j is the imaginary unit. The Doppler measurement unit 40 obtains the autocorrelation value A(r) for N received Doppler signals z(r, 1), z(r, 2), ···· z(r, N) according to (Equation 2).

[0037]

Number

[0038] Here, the superscript "*" indicates that it is a conjugate complex number. The Doppler measurement unit 40 obtains the blood flow velocity v(r) according to (Equation 3) using the real part Re[A(r)] and the imaginary part Im[A(r)] of the autocorrelation value A(r).

[0039]

Number

[0040] c is the propagation speed of the ultrasonic pulse propagating in the subject 18, f0 is the frequency of the ultrasonic pulse transmitted from the ultrasonic probe 14. T is the time interval for transmitting N ultrasonic pulses. The part on the right side of c / (2f0) in the right side of (Equation 3) represents the Doppler frequency. The blood flow velocity v(r) represents the depth direction component of the blood flow velocity at the depth r. When the angle φ formed by the direction of the blood flow with respect to the depth direction is known, v(r) / cosφ is the blood flow velocity.

[0041] When measuring the blood flow velocity distribution or the like, when it is not necessary to obtain the absolute value of the blood flow velocity (for example, the value in the MKSA unit system), the blood flow velocity may be a value obtained by multiplying the value obtained by (Equation 3) by an arbitrary constant. Further, instead of the blood flow velocity, the Doppler shift frequency measurement value, which is the part on the right side of the right side of (Equation 3) than c / (2f0), may be used.

[0042] The image processing unit 34 generates data indicating a B-mode color Doppler image in which a color corresponding to the blood flow velocity is added to the B-mode image based on the B-mode image data and the color Doppler data. The image processing unit 34 generates a video signal based on the B-mode color Doppler image data and outputs it to the display unit 46. The display unit 46 displays the B-mode color Doppler image based on the video signal.

[0043] The Doppler measurement unit 40 may use the autocorrelation value A(r) at each depth r obtained for each of a plurality of transmission / reception beams having different positions or directions as one piece of Doppler measurement information.

[0044] Further, the Doppler measurement unit 40 may use the velocity v(r) at each depth r obtained for each of a plurality of transmission / reception beams having different positions or directions as one piece of Doppler measurement information.

[0045] The Doppler measurement unit 40 may generate the blood flow power P(r) at each depth r obtained for each of the transmission / reception beams having different positions or directions as one piece of Doppler measurement information. The blood flow power P(r) is represented by the magnitude of the received Doppler signal represented by (Equation 1). In order for the Doppler measurement unit 40 to obtain the blood flow power P(r), an appropriate filter such as an SVD (Singular Value Decomposition) filter may be used as the wall filter 38.

[0046] Further, the Doppler measurement unit 40 may generate the degree of variation in the Doppler frequency of the received Doppler signals acquired N times for one transmission / reception beam as one piece of information included in the Doppler measurement information (hereinafter referred to as the degree of Doppler frequency variation).

[0047] The Doppler frequency variation degree may be obtained based on N - 1 adjacent time Doppler frequencies obtained according to (Equation 4) when the blood flow velocity v(r) is obtained according to the above (Equation 1) to (Equation 3).

[0048] That is, the Doppler measurement unit 40 uses the real part and the imaginary part of z(r,q)·z(r,q + 1) that are added and summed on the right side of (Equation 2) to obtain the adjacent time Doppler frequency fa(r,q) for each of q = 1 to N - 1 according to (Equation 4). * The Doppler measurement unit 40 obtains the adjacent time Doppler frequencies fa(r,q) for each of q = 1 to N - 1 according to (Equation 4) using the real part and the imaginary part of z(r,q)·z(r,q + 1).

[0049]

Equation

[0050] The Doppler measurement unit 40 obtains the Doppler frequency variation degree for N - 1 adjacent time Doppler frequencies fa(r,1), fa(r,2), ··· fa(r,N - 1) and outputs it to the image processing unit 34. The Doppler frequency variation degree may be a value indicating the variation such as the variance and the standard deviation for N - 1 adjacent time Doppler frequencies.

[0051] The image processing unit 34 may adjust the color attached to the B - mode image according to the Doppler frequency variation degree. For example, when blue is attached to the region where blood flows in the direction away from the ultrasonic probe 14 and red is attached to the region where blood flows in the direction approaching the ultrasonic probe 14, the larger the Doppler frequency variation degree, the closer the color may be to purple or white.

[0052] As described above, the Doppler measurement information generated by the Doppler measurement unit 40 includes, in addition to the color Doppler data, the autocorrelation value A(r), the blood flow velocity v(r), the blood flow power P(r), the Doppler frequency variation degree, etc. obtained for each of the transmission - reception beams with different positions or directions. The Doppler measurement information may be information including at least one of the autocorrelation value A(r), the blood flow velocity v(r), the blood flow power P(r), and the Doppler frequency variation degree.

[0053] Thus, in the basic operation, based on the N received Doppler signals that have been subjected to high-pass filtering by the wall filter 38, the Doppler measurement unit 40 generates Doppler measurement information. The received Doppler signals output from the quadrature detection unit 36 contain low-frequency noise based on the movement of the tissue within the subject 18, which causes unnecessary components included in the Doppler measurement information. For example, for color Doppler data, it causes noise called clutter. Therefore, the unnecessary components included in the Doppler measurement information are suppressed by the wall filter 38.

[0054] Next, the learning operation will be described. In the learning operation, the following processing is executed when ultrasonic pulses are transmitted and received K times for each transmission / reception beam. Here, K is an integer of 2 or more that is smaller than the above-mentioned N. With the ultrasonic pulses being transmitted and received K times for each transmission / reception beam, K received Doppler signals are output from the quadrature detection unit 36 to the machine learning unit 42 and the wall filter 38.

[0055] The wall filter 38 performs high-pass filtering on the K received Doppler signals and outputs the K received Doppler signals after the high-pass filtering to the Doppler measurement unit 40 and the machine learning unit 42. The Doppler measurement unit 40 generates Doppler measurement information based on the received Doppler signals acquired K times for each of a plurality of transmission / reception beams having different positions or directions, and outputs it to the machine learning unit 42.

[0056] Thus, the following information is input as training information to the machine learning unit 42 when ultrasonic pulses are transmitted and received K times (hereinafter sometimes referred to as K times of transmission / reception) for each transmission / reception beam. That is, the received Doppler signal before being subjected to high-pass filtering by the wall filter 38, the received Doppler signal after being subjected to high-pass filtering by the wall filter 38, and the Doppler measurement information generated by the Doppler measurement unit 40 are input to the machine learning unit 42.

[0057] The training information may include at least one of the received Doppler signal before high-pass filtering by the wall filter 38, the received Doppler signal after high-pass filtering by the wall filter 38, and the Doppler measurement information generated by the Doppler measurement unit 40. The Doppler measurement information may be at least one of a plurality of types of information included in the Doppler measurement information.

[0058] Next, when ultrasonic pulses are transmitted and received J times (hereinafter sometimes referred to as J times of transmission and reception) for each transmission / reception beam, the following information is input to the machine learning unit 42 as target information. Here, J is an integer of 2 or more that is larger than K. In the case of J times of transmission and reception, the tap coefficients of the wall filter 38 are larger than in the case of K times of transmission and reception. Therefore, in the case of J times of transmission and reception, compared with the case of K times of transmission and reception, the low-pass filter characteristics become steeper, and the characteristics of small attenuation in the high frequency range and large attenuation in the low frequency range become prominent.

[0059] The received Doppler signal before high-pass filtering by the wall filter 38, the received Doppler signal after high-pass filtering by the wall filter 38, and further, the Doppler measurement information generated by the Doppler measurement unit 40 are input to the machine learning unit 42. However, the region (measurement object) where ultrasonic pulses are transmitted and received in the case of J times of transmission and reception is the same as the region (measurement object) where ultrasonic pulses are transmitted and received in the case of K times of transmission and reception. The Doppler measurement information as the target information may be at least one of a plurality of types of information included in the Doppler measurement information.

[0060] In the learning operation, the above-described K transmissions and receptions and J transmissions and receptions are performed on a plurality of different measurement objects, for example, blood flows at a plurality of different observation sites. The machine learning unit 42 acquires, as teacher data, a set of training information generated by the K transmissions and receptions and target information generated by the J transmissions and receptions for a plurality of different subjects 18 and a plurality of different observation sites, and constructs a machine learning model 44. For at least one of the plurality of different measurement objects, the target information includes at least Doppler measurement information. The learning operation may be performed using a machine learning algorithm such as CNN, AdaBoost, or SVN. The machine learning algorithm may include a deep learning algorithm.

[0061] The constructed machine learning model 44 is a model that generates Doppler measurement information equivalent to that in the case of J transmissions and receptions when the following input information is input when ultrasonic pulses are transmitted and received K times for each transmission and reception beam. That is, the machine learning model 44 is a model that predicts Doppler measurement information that would be obtained if ultrasonic pulses were transmitted and received J times when ultrasonic pulses are transmitted and received K times.

[0062] The input information may be at least one of a received Doppler signal before high-pass filtering is performed by the wall filter 38, a received Doppler signal after high-pass filtering is performed by the wall filter 38, and Doppler measurement information generated by the Doppler measurement unit 40. The Doppler measurement information may be at least one of a plurality of types of information included in the Doppler measurement information.

[0063] Next, the model application operation will be described. In the model application operation, an ultrasonic beam is scanned at an observation site of the subject 18, and ultrasonic pulses are transmitted and received for a plurality of transmission and reception beams having different positions or directions. For each transmission and reception beam, ultrasonic pulses are transmitted and received K times, and K received Doppler signals are output from the quadrature detection unit 36 to the wall filter 38 and the machine learning model 44.

[0064] The wall filter 38 performs high-pass filtering on the K received Doppler signals and outputs the K received Doppler signals after the high-pass filtering to the Doppler measurement unit 40 and the machine learning model 44. The Doppler measurement unit 40 generates Doppler measurement information based on the received Doppler signals acquired K times for each of a plurality of transmission / reception beams having different positions or directions, and outputs the information to the machine learning model 44.

[0065] As a result, input information is input to the machine learning model 44 for the case where ultrasonic pulses are transmitted and received K times for each transmission / reception beam, and the machine learning model 44 generates Doppler measurement information equivalent to the case of J transmissions and receptions for the input information.

[0066] When the Doppler measurement information includes color Doppler data, the machine learning model 44 outputs the color Doppler data to the image processing unit 34. The image processing unit 34 generates data indicating a B-mode color Doppler image based on the B-mode image data and the color Doppler data. The image processing unit 34 generates a video signal for displaying the B-mode color Doppler image based on the B-mode color Doppler image data, and outputs the video signal to the display unit 46. The display unit 46 displays the B-mode color Doppler image based on the video signal.

[0067] Note that the image processing unit 34 may generate composite image data for displaying side by side an image based on the B-color Doppler image data obtained by the normal operation and an image based on the B-color Doppler image data obtained by the model application operation. Further, the image processing unit 34 may generate composite image data indicating an image in which one of the image based on the B-color Doppler image data obtained by the normal operation and the image based on the B-color Doppler image data obtained by the model application operation is seen through the other. The image processing unit 34 generates a video signal for displaying the composite image based on the composite image data, and outputs the video signal to the display unit 46. The display unit 46 displays the composite image based on the video signal.

[0068] As described above, the ultrasonic diagnostic apparatus 100 according to this embodiment includes a transmission unit 12, a reception unit 20, and an information processing unit 26. The transmission unit 12 transmits ultrasonic pulses N times by the ultrasonic probe 14. The reception unit 20 receives the reflected waves generated N times in the measurement object by the ultrasonic probe 14. The information processing unit 26 generates N reception Doppler signals from the N reception pulse signals output from the reception unit 20 by the quadrature detection unit 36 according to the reflected waves generated N times, and executes processing on the N reception Dopplers. The measurement object in this embodiment is an observation site where blood is flowing in the subject 18.

[0069] The information processing unit 26 includes a wall filter 38 (filter) that performs high-pass filter processing on the N reception Doppler signals, a Doppler measurement unit 40 that generates Doppler measurement information of the measurement object based on the N reception Doppler signals subjected to high-pass filter processing, and a machine learning model 44 constructed based on the teacher data.

[0070] The teacher data includes training information and target information. The training information is at least one of reception characteristic information derived from each reception Doppler signal with N being K, and Doppler measurement information generated by the Doppler measurement unit 40 when N is K. Here, the reception characteristic information includes at least one of the reception Doppler signal before high-pass filter processing and the reception Doppler signal after high-pass filter processing.

[0071] The target information is at least one of reception characteristic information derived from each reception Doppler signal with N being J for the same measurement object as when N is K, and Doppler measurement information generated by the Doppler measurement unit 40 when N is J. Here, J and K are integers of 2 or more, and J is larger than K.

[0072] The machine learning model 44 generates Doppler measurement information based on the input information. The input information is at least one of the reception characteristic information derived from each received Doppler signal with N being K for the measurement object in the subject 18, and the Doppler measurement information generated by the Doppler measurement unit 40 when N is K.

[0073] As described above, in the case of J transmissions and receptions with N being J, compared with the case of K transmissions and receptions with N being K, the tap coefficients of the wall filter 38 increase. Therefore, in the case of J transmissions and receptions, compared with the case of K transmissions and receptions, the low-pass filter characteristics become steeper, and the characteristics that the attenuation amount is small in the high frequency range and large in the low frequency range become prominent. Therefore, in the case of J transmissions and receptions, the unnecessary components included in the Doppler measurement information are more greatly suppressed than in the case of K transmissions and receptions. Regarding the color Doppler data included in the Doppler measurement information, in the case of J transmissions and receptions, the effect of suppressing clutter is greater than in the case of K transmissions and receptions.

[0074] According to the ultrasonic diagnostic apparatus 100 according to the present embodiment, by applying the machine learning model 44, Doppler measurement information equivalent to that in the case of J transmissions and receptions can be obtained even in the case of K transmissions and receptions. As a result, even when the number of transmissions of the ultrasonic pulse is relatively small, unnecessary components included in the Doppler measurement information are suppressed by the wall filter 38. In the case of displaying a B-mode color Doppler image, even when the number of transmissions of the ultrasonic pulse is relatively small, clutter is suppressed by the wall filter 38.

[0075] Also, in the ultrasonic diagnostic apparatus 100 according to the present embodiment, by reducing the number of transmissions of the ultrasonic pulse while maintaining the effect of suppressing unnecessary components included in the Doppler measurement information, the frame rate when generating color Doppler data increases.

[0076] FIG. 2 shows the operation in which training information is input to the machine learning unit 42. This figure shows that J = 3K holds, and for J transmissions and receptions of a series of ultrasonic pulses, the training information is input to the machine learning unit 42 three times. That is, as J ultrasonic pulses are transmitted from the ultrasonic probe 14 and J reflected waves are received by the ultrasonic probe 14, J received Doppler signals are input to the wall filter 38 and the machine learning unit 42.

[0077] The wall filter 38, the Doppler measurement unit 40, and the machine learning unit 42 input training information to the machine learning unit 42 for every K received Doppler signals. That is, every time K received Doppler signals are input, the wall filter 38 outputs the K received Doppler signals after high-pass filtering to the Doppler measurement unit 40 and the machine learning unit 42. Every time the K received Doppler signals after high-pass filtering are input, the Doppler measurement unit 40 outputs Doppler measurement information to the machine learning unit 42. In this way, by inputting K received Doppler signals, K received Doppler signals before high-pass filtering, K received Doppler signals after high-pass filtering, and Doppler measurement information generated according to the K received Doppler signals, the training information corresponding to K transmissions and receptions is input to the machine learning unit 42 three times.

[0078] FIG. 3 shows the operation in which target information is input to the machine learning unit 42. This figure shows that for J transmissions and receptions of a series of ultrasonic pulses, the target information is input to the machine learning unit 42 once. That is, as J ultrasonic pulses are transmitted from the ultrasonic probe 14 and J reflected waves are received by the ultrasonic probe 14, J received Doppler signals are input to the wall filter 38 and the machine learning unit 42. The J received Doppler signals shown in FIG. 3 may be the same as the J received Doppler signals shown in FIG. 2. That is, the J received Doppler signals obtained by J transmissions and receptions in the operation in which the training information is input to the machine learning unit 42 may be used in the operation in which the target information is input to the machine learning unit 42.

[0079] The wall filter 38, the Doppler measurement unit 40, and the machine learning unit 42 input target information based on the J received Doppler signals into the machine learning unit 42. That is, the wall filter 38 outputs the J received Doppler signals after high-pass filtering to the Doppler measurement unit 40 and the machine learning unit 42 based on the J received Doppler signals. In response to the input of the J received Doppler signals after high-pass filtering, the Doppler measurement unit 40 outputs Doppler measurement information to the machine learning unit 42. In this way, the J received Doppler signals, the J received Doppler signals before high-pass filtering, the J received Doppler signals after high-pass filtering, and the Doppler measurement information generated according to the J received Doppler signals are used as target information corresponding to J transmissions and receptions are input into the machine learning unit 42.

[0080] The operations shown in FIGS. 2 and 3 are performed on a plurality of different subjects 18 and a plurality of different observation sites, and the machine learning unit 42 constructs a machine learning model 44 based on the training information input multiple times and the target information input multiple times.

[0081] Model application operation is shown in FIG. 4. In this figure, for a series of K ultrasonic transmissions and receptions, K received Doppler signals before high-pass filtering, K received Doppler signals after high-pass filtering, and Doppler measurement information are shown to be input once into the machine learning model 44. That is, as K ultrasonic pulses are transmitted from the ultrasonic probe 14 and K reflected waves are received by the ultrasonic probe 14, the K received Doppler signals are input into the wall filter 38 and the machine learning model 44. Based on the K received Doppler signals, the wall filter 38 outputs the K received Doppler signals after high-pass filtering to the Doppler measurement unit 40 and the machine learning model 44. The Doppler measurement unit 40 outputs Doppler measurement information to the machine learning model 44 when the K received Doppler signals after high-pass filtering are input. In this way, the K received Doppler signals before high-pass filtering, the K received Doppler signals after high-pass filtering, and the Doppler measurement information generated according to the K received Doppler signals are input as input information into the machine learning model 48, and the machine learning model 48 generates Doppler measurement information equivalent to that in the case of J transmissions and receptions based on the input information.

[0082] FIG. 5 shows an example of ultrasonic pulses transmitted from the ultrasonic probe 14 in normal operation. A plurality of vibration elements 16 provided in the ultrasonic probe 14 are arranged in the left-right direction of FIG. 5. The vertical axis represents time. In the embodiment shown in FIG. 5, four transmission bands TX1 to TX4 are formed at different positions in the horizontal direction by the plurality of vibration elements 16 arranged in the left-right direction. Each transmission band TX1 to TX4 is formed by a plurality of transmission beams. In each transmission band TX1 to TX4, N ultrasonic pulses are transmitted, and N reflected waves generated by reflection from the subject 18 are received. A series of N ultrasonic pulses is regarded as an ultrasonic pulse group, and the ultrasonic pulse group is transmitted at a predetermined time interval.

[0083] In adjacent transmission bands TX1 and TX2, ultrasonic pulses are transmitted alternately, and N ultrasonic pulses are transmitted in each of the transmission bands TX1 and TX2. Similarly, in adjacent transmission bands TX3 and TX4, ultrasonic pulses are transmitted alternately, and N ultrasonic pulses are transmitted in each of the transmission bands TX3 and TX4.

[0084] After an ultrasonic pulse group SP1 is transmitted in adjacent transmission bands TX1 and TX2, an ultrasonic pulse group SP2 is transmitted in adjacent transmission bands TX3 and TX4. The transmission of the ultrasonic pulse group SP1 in the adjacent transmission bands TX1 and TX2 and the transmission of the ultrasonic pulse group SP2 in the adjacent transmission bands TX3 and TX4 are repeated alternately. In the following description, the adjacent transmission bands TX1 and TX2 are referred to as a transmission band pair PTX1, and the adjacent transmission bands TX3 and TX4 are referred to as a transmission band pair PTX2.

[0085] After the ultrasonic pulse group SP2 is transmitted in the transmission band pair PTX2 and before the ultrasonic pulse group SP2 is transmitted in the transmission band pair PTX1, in the ultrasonic probe 14, ultrasonic pulses for B-mode image generation are transmitted, and reflected waves for B-mode image generation are received. In the embodiment shown in FIG. 5, a half transmission band for B-mode image generation is formed with a width that is half of the width of each of the transmission bands TX1 to TX4. Two ultrasonic pulses for B-mode image generation are transmitted in each of two adjacent half transmission bands within the transmission band TX1, and the ultrasonic pulses for B-mode image generation are transmitted alternately in two adjacent half transmission bands within the transmission band TX1. In this embodiment, the B mode is a phase inversion B mode, and the phase difference between the two ultrasonic pulses for B-mode image generation transmitted in each half transmission band is 180°.

[0086] In the present embodiment, an operation is performed in which ultrasonic pulses for generating a B-mode image are alternately transmitted in two half-transmission bands in each transmission band in the order of transmission bands TX1, TX2, TX2, and TX4. In FIG. 5, an ultrasonic pulse group for a B-mode image when ultrasonic pulses for generating a B-mode image are alternately transmitted in two half-transmission bands in each of the transmission bands TX1, TX2, TX3, and TX4 is shown as an ultrasonic pulse group PIB.

[0087] An ultrasonic pulse group SP1 is transmitted in a transmission band pair PTX1, an ultrasonic pulse group SP2 is transmitted in a transmission band pair PTX2, an ultrasonic pulse group PIB is transmitted in each of the transmission bands TX1, TX2, TX2, and TX4, and processing for one frame FR is executed by reception processing for the reflected waves generated by each ultrasonic pulse group. The processing for one frame FR corresponds to processing for generating Doppler measurement information corresponding to one color Doppler image.

[0088] For each of the transmission bands TX1 to TX4, in the processing for one frame FR, when N ultrasonic pulses are transmitted and N reflected waves generated by reflection in the subject 18 are received, a normal operation is executed, and the Doppler measurement unit 40 generates Doppler measurement information not depending on the machine learning model 44.

[0089] FIG. 6 shows an example of ultrasonic pulses transmitted from the ultrasonic probe 14 in the model application operation. Also in the embodiment shown in FIG. 6, similar to the embodiment shown in FIG. 5, four transmission bands TX1 to TX4 are formed at different positions in the lateral direction for the ultrasonic pulses by a plurality of vibration elements 16 arranged in the left-right direction.

[0090] In each of the adjacent transmission bands TX1 and TX2, two ultrasonic pulses are transmitted, and the ultrasonic pulses are transmitted alternately in the adjacent transmission bands TX1 and TX2. Similarly, in each of the adjacent transmission bands TX3 and TX4, two ultrasonic pulses are transmitted, and the ultrasonic pulses are transmitted alternately in the adjacent transmission bands TX3 and TX4. In this way, in each of the transmission bands TX1, TX2, TX3, and TX4, an ultrasonic pulse group SP1 consisting of two ultrasonic pulses is transmitted. Similarly, in each of the transmission bands TX1, TX2, TX3, and TX4, ultrasonic pulse groups SP2, SP3, and SP4, each consisting of two ultrasonic pulses, are transmitted in order over time.

[0091] In this embodiment, in each of the transmission bands TX1, TX2, TX3, and TX4, a series of transmission processes in which the ultrasonic pulse groups SP1 to SP4 are transmitted in order over time is repeated. By transmitting the ultrasonic pulse groups SP1 to SP4 in each of the transmission bands TX1, TX2, TX3, and TX4, K ultrasonic pulses are transmitted in each of the transmission bands TX1, TX2, TX3, and TX4. In the example shown in FIG. 6, K = 8.

[0092] The transmission of the ultrasonic waves for generating the B-mode image transmitted along with the transmission of the first series of ultrasonic pulse groups SP1 to SP4 in FIG. 6 will be described. While the ultrasonic pulse group SP1 is being transmitted in the transmission band TX4 until the ultrasonic pulse group SP2 is transmitted in the transmission band TX1, an ultrasonic pulse group PIB1 for generating the B-mode image is transmitted in the half transmission band on the left side of the transmission band TX1. While the ultrasonic pulse group SP2 is being transmitted in the transmission band TX4 until the ultrasonic pulse group SP3 is transmitted in the transmission band TX1, an ultrasonic pulse group PIB2 for generating the B-mode image is transmitted in the half transmission band on the right side of the transmission band TX1.

[0093] After the ultrasonic pulse group SP3 is transmitted in the transmission band TX4 and before the ultrasonic pulse group SP4 is transmitted in the transmission band TX1, the ultrasonic pulse group PIB3 for B-mode image generation is transmitted in the left half transmission band of the transmission band TX2. After the ultrasonic pulse group SP4 is transmitted in the transmission band TX4 and before the ultrasonic pulse group SP1 is transmitted in the transmission band TX1, the ultrasonic pulse group PIB4 for B-mode image generation is transmitted in the right half transmission band of the transmission band TX2.

[0094] The transmission of the ultrasonic waves for B-mode image generation transmitted along with the transmission of the second series of ultrasonic pulse groups SP1 to SP4 in FIG. 6 will be described. After the ultrasonic pulse group SP1 is transmitted in the transmission band TX4 and before the ultrasonic pulse group SP2 is transmitted in the transmission band TX1, the ultrasonic pulse group PIB5 for B-mode image generation is transmitted in the left half transmission band of the transmission band TX3. After the ultrasonic pulse group SP2 is transmitted in the transmission band TX4 and before the ultrasonic pulse group SP3 is transmitted in the transmission band TX1, the ultrasonic pulse group PIB6 for B-mode image generation is transmitted in the right half transmission band of the transmission band TX3. After the ultrasonic pulse group SP3 is transmitted in the transmission band TX4 and before the ultrasonic pulse group SP4 is transmitted in the transmission band TX1, the ultrasonic pulse group PIB7 for B-mode image generation is transmitted in the left half transmission band of the transmission band TX4. After the ultrasonic pulse group SP4 is transmitted in the transmission band TX4 and before the ultrasonic pulse group SP1 is transmitted in the transmission band TX1, the ultrasonic pulse group PIB8 for B-mode image generation is transmitted in the right half transmission band of the transmission band TX4.

[0095] Thereafter, along with the transmission of the odd-numbered series of ultrasonic pulse groups SP1 to SP4 in FIG. 6, the ultrasonic pulse groups PIB1 to PIB4 are transmitted at the same timing as the ultrasonic pulse groups PIB1 to PIB4 transmitted along with the transmission of the first series of ultrasonic pulse groups SP1 to SP4. Also, along with the transmission of the even-numbered series of ultrasonic pulse groups SP1 to SP4 in FIG. 6, the ultrasonic pulse groups PIB5 to PIB8 are transmitted at the same timing as the ultrasonic pulse groups PIB5 to PIB8 transmitted along with the transmission of the second series of ultrasonic pulse groups SP1 to SP4.

[0096] In each of the transmission bands TX1 to TX4, ultrasonic pulse groups SP1 to SP4 are transmitted, and by performing reception processing on the reflected waves generated by each ultrasonic pulse group, processing for one frame FR of color Doppler data is executed. The processing for one frame FR corresponds to processing for generating Doppler measurement information corresponding to one color Doppler image.

[0097] Also, from the half transmission bands formed by each of the transmission bands TX1 to TX4, ultrasonic pulse groups PIB1 to PIB8 for generating a B-mode image are transmitted, and by performing reception processing on the reflected waves generated by each ultrasonic pulse group, processing for one frame FRB of phase inversion B-mode data is executed. The processing for one frame FRB corresponds to processing for generating phase inversion B-mode data corresponding to one phase inversion B-mode image.

[0098] In the transmission of ultrasonic pulses in the normal operation shown in FIG. 5, while generating Doppler measurement information for one frame FR, for two adjacent transmission band pairs PTX1 and PTX2, the operation of alternately transmitting ultrasonic pulse groups is performed only once.

[0099] On the other hand, in the model application operation shown in FIG. 6, while generating Doppler measurement information for one frame FR, for two adjacent transmission band pairs PTX1 and PTX2, the operation of alternately transmitting ultrasonic pulse groups is performed a plurality of times. In the example shown in FIG. 6, while generating Doppler measurement information for one frame FR, for two adjacent transmission band pairs PTX1 and PTX2, the operation of alternately transmitting ultrasonic pulse groups is performed four times.

[0100] Therefore, compared with FIG. 5, the time interval during which the ultrasonic pulse group is transmitted in one frame FR becomes larger, and the phase change of a plurality of received Doppler signals obtained in one frame FR becomes sufficiently large. That is, the phase angle (argument of a complex number) of the autocorrelation value A(r) represented by (Equation 2) becomes sufficiently large. As a result, the measurement accuracy of Doppler measurement information is improved. Further, for two adjacent transmission band pairs PTX1 and PTX2, by alternately transmitting the ultrasonic pulse groups, the area where the Doppler measurement information is acquired per unit time becomes wider. Further, compared with FIG. 5, the time required to generate the Doppler measurement information for one frame FR becomes shorter, and the frame rate becomes higher.

[0101] Here, an embodiment is shown in which the ultrasonic probe 14 includes a plurality of vibrating elements 16 arranged linearly, and a plurality of transmission bands are formed in a direction perpendicular to the arrangement direction of the vibrating elements 16. The plurality of vibrating elements 16 may be arranged in a curved shape at the tip of the ultrasonic probe 14. In this case, a plurality of transmission bands may be formed so as to extend radially from the tip of the ultrasonic probe 14.

[0102] FIG. 7 shows a model application operation when an ultrasonic pulse is transmitted from the ultrasonic probe 14 according to FIG. 6. In this figure, for a series of K transmissions and receptions of ultrasonic waves for one frame FR, K received Doppler signals before high-pass filtering, K received Doppler signals after high-pass filtering, and Doppler measurement information are shown to be input once into the machine learning model 44. That is, as K ultrasonic pulses are transmitted from the ultrasonic probe 14 and K reflected waves are received by the ultrasonic probe 14, K received Doppler signals are input to the wall filter 38 and the machine learning model 44. The wall filter 38 outputs the K received Doppler signals after high-pass filtering to the Doppler measurement unit 40 and the machine learning model 44. The Doppler measurement unit 40 outputs Doppler measurement information to the machine learning model 44 when the K received Doppler signals after high-pass filtering are input. In this way, the K received Doppler signals, the K received Doppler signals before high-pass filtering, the K received Doppler signals after high-pass filtering, and the Doppler measurement information generated according to the K received Doppler signals are input to the machine learning model 44 as input information. The machine learning model 44 generates Doppler measurement information equivalent to that in the case of J transmissions and receptions based on the input information. For at least one of the time intervals between each adjacent received pulse signal in the K received Doppler signals, a time interval in which the Doppler frequency can be detected may be used.

[0103] Note that the ultrasonic pulse group PIB shown in FIG. 5 and the ultrasonic pulse groups PIB1 to PIB8 shown in FIG. 6 may be used as ultrasonic pulse groups for the Doppler mode in addition to the B mode. In this case, according to the transmission and reception of the ultrasonic pulses belonging to the ultrasonic pulse group PIB or the ultrasonic pulse groups PIB1 to PIB8, the received pulse signal output from the coherent addition unit 30 is output not only to the B mode image generation unit 32 but also to the quadrature detection unit 36. For example, by using ultrasonic pulses having the same waveform for both the B mode and the Doppler mode, the ultrasonic pulses in the B mode can be used in the Doppler mode. For ultrasonic pulses using phase inversion, the quadrature detection unit 36 gives a 180° phase difference to two received Doppler signals based on two ultrasonic pulses.

[0104] According to the reflected wave for B mode image generation, the received pulse signal as the coherently received signal (B mode image received signal) output from the receiving unit 20 is used as one of the K received pulse signals output from the receiving unit 20 for the operation in the Doppler mode. By using the ultrasonic pulse group PIB or the ultrasonic pulse groups PIB1 to PIB8 as the ultrasonic pulse group for the Doppler mode in addition to the B mode, an effect equivalent to the case where the number of transmissions and receptions of the transmission and reception pulses is increased in the operation of the Doppler mode can be obtained.

[0105] [Configuration of the Present Invention] Configuration 1: Let N be an integer of 2 or more. A transmission unit that transmits ultrasonic pulses N times by an ultrasonic probe. A receiving unit that receives the reflected waves generated N times by the measurement object by the ultrasonic probe. An information processing unit that generates N received Doppler signals from N received pulse signals output from the receiving unit according to the reflected waves generated N times by the measurement object and executes processing for each of the received Doppler signals. The information processing unit A filter that performs high-pass filter processing on the N received Doppler signals. A Doppler measurement unit that generates Doppler measurement information of a measurement object based on the N received Doppler signals subjected to the high-pass filter processing; including a machine learning model constructed based on teacher data; Let J and K be integers greater than or equal to 2, and under the condition that J is greater than K, The teacher data is reception characteristic information derived from each of the received Doppler signals with N being K, and training information that is at least one of the Doppler measurement information generated by the Doppler measurement unit when N is K; For the same measurement object as when N is K, reception characteristic information derived from each of the received Doppler signals with N being J, and target information that is at least one of the Doppler measurement information generated by the Doppler measurement unit when N is J; The machine learning model is An ultrasonic diagnostic apparatus characterized by generating the Doppler measurement information based on the reception characteristic information derived from each of the received Doppler signals with N being K, and input information that is at least one of the Doppler measurement information generated by the Doppler measurement unit when N is K, for a measurement object in a subject. Configuration 2: The ultrasonic diagnostic apparatus according to Configuration 1, wherein the reception characteristic information is the N received Doppler signals before the high-pass filter processing, and at least one of the N received Doppler signals after the high-pass filter processing. Configuration 3: The ultrasonic diagnostic apparatus according to Configuration 1 or Configuration 2, wherein the Doppler measurement information is the autocorrelation value of the N received Doppler signals after the high-pass filter processing, the velocity of the measurement object obtained from the N received Doppler signals after the high-pass filter processing, the Doppler frequency variation degree for the N received Doppler signals after the high-pass filter processing, and An ultrasonic diagnostic apparatus, characterized by including at least one of values indicating the magnitudes of the N received Doppler signals after the high-pass filtering process. Configuration 4: The ultrasonic diagnostic apparatus according to any one of Configurations 1 to 3, wherein the transmission unit transmits ultrasonic waves for generating a B-mode image by the ultrasonic probe, the reception unit receives, by the ultrasonic probe, reflected waves for generating a B-mode image generated in the measurement object, and the information processing unit includes a B-mode image generation unit that generates B-mode image data based on a B-mode image reception signal output from the reception unit in accordance with the reflected waves for generating a B-mode image, An ultrasonic diagnostic apparatus, characterized in that the B-mode image reception signal output from the reception unit is used as any one of the N reception pulse signals output from the reception unit in accordance with the reflected waves for generating a B-mode image. Configuration 5: The ultrasonic diagnostic apparatus according to any one of Configurations 1 to 4, wherein the transmission unit forms a plurality of transmission bands at different positions or in different directions for the ultrasonic pulses, and the ultrasonic pulses are transmitted alternately such that, for two adjacent transmission bands, while the ultrasonic pulses are not being transmitted in one transmission band, the ultrasonic pulses are transmitted in the other transmission band, and for two adjacent transmission band pairs each including two adjacent transmission bands, a series of the ultrasonic pulses transmitted alternately in the two transmission bands included in one transmission band pair is defined as an ultrasonic pulse group, and the ultrasonic pulse groups are transmitted alternately such that, while the ultrasonic pulse groups are not being transmitted in one transmission band pair, the ultrasonic pulse groups are transmitted in the other transmission band pair, An ultrasonic diagnostic apparatus, characterized in that each time K ultrasonic pulses are transmitted in each transmission band, the machine learning model generates Doppler measurement information for one frame for each transmission band. Configuration 6: The ultrasonic diagnostic apparatus according to Configuration 5, During generation of the Doppler measurement information for one frame, an operation in which ultrasonic pulse groups are alternately transmitted for two adjacent transmission band pairs is performed a plurality of times. An ultrasonic diagnostic apparatus characterized by this.

Explanation of Signs

[0106] 10 beam control unit, 12 transmission unit, 14 ultrasonic probe, 16 vibration element, 18 subject, 20 reception unit, 22 control unit, 24 operation unit, 26 information processing unit, 30 coherent addition unit, 32 B-mode image generation unit, 34 image processing unit, 36 quadrature detection unit, 38 wall filter, 40 Doppler measurement unit, 42 machine learning unit, 44 machine learning model, 46 display unit, 100 ultrasonic diagnostic apparatus.

Claims

1. Let N be an integer of 2 or more, a transmission unit that transmits ultrasonic pulses N times by an ultrasonic probe; a reception unit that receives reflected waves generated N times in the measurement object by the ultrasonic probe; an information processing unit that generates N reception Doppler signals from N reception pulse signals output from the reception unit in response to the reflected waves generated N times in the measurement object, and executes processing for each of the reception Dopplers, comprising: the information processing unit a filter that performs high-pass filter processing on the N reception Doppler signals; a Doppler measurement unit that generates Doppler measurement information of the measurement object based on the N reception Doppler signals subjected to the high-pass filter processing; including a machine learning model constructed based on teacher data; Let J and K be integers of 2 or more, and under the condition that J is greater than K, the teacher data reception characteristic information derived from each of the reception Doppler signals with N being K, and training information that is at least one of the Doppler measurement information generated by the Doppler measurement unit when N is K; for the same measurement object as when N is K, the reception characteristic information derived from each of the reception Doppler signals with N being J, and target information that is at least one of the Doppler measurement information generated by the Doppler measurement unit when N is J; the machine learning model An ultrasonic diagnostic apparatus, characterized in that, for a measurement object in a subject, based on input information that is at least one of the reception characteristic information derived from each of the reception Doppler signals with N being K, and the Doppler measurement information generated by the Doppler measurement unit when N is K, the Doppler measurement information is generated.

2. The ultrasonic diagnostic apparatus according to claim 1, wherein the reception characteristic information includes at least one of the N reception Doppler signals before the high-pass filter processing and at least one of the N reception Doppler signals after the high-pass filter processing.

3. The ultrasonic diagnostic apparatus according to claim 1, wherein the Doppler measurement information is the autocorrelation value of the N reception Doppler signals after the high-pass filter processing, the velocity of the measurement object obtained from the N reception Doppler signals after the high-pass filter processing, the degree of Doppler frequency variation for the N reception Doppler signals after the high-pass filter processing, and An ultrasonic diagnostic apparatus, characterized by including at least one of values indicating the magnitudes of the N received Doppler signals after the high-pass filter processing.

4. The ultrasonic diagnostic apparatus according to any one of Claims 1 to 3, wherein the transmission unit transmits ultrasonic waves for B-mode image generation by the ultrasonic probe, the reception unit receives reflected waves for B-mode image generation generated in the measurement object by the ultrasonic probe, the information processing unit includes a B-mode image generation unit that generates B-mode image data based on a B-mode image reception signal output from the reception unit in accordance with the reflected waves for B-mode image generation, An ultrasonic diagnostic apparatus, characterized in that the B-mode image reception signal output from the reception unit is used as any one of the N received pulse signals output from the reception unit in accordance with the reflected waves for B-mode image generation.

5. The ultrasonic diagnostic apparatus according to any one of Claims 1 to 3, wherein the transmission unit forms a plurality of transmission bands at different positions or in different directions for the ultrasonic pulse, and the ultrasonic pulse is transmitted alternately so that the ultrasonic pulse is transmitted in one transmission band while the ultrasonic pulse is not transmitted in the other transmission band for two adjacent transmission bands. For two adjacent transmission band pairs each including two adjacent transmission bands, a series of the ultrasonic pulses transmitted alternately in the two transmission bands included in one transmission band pair is defined as an ultrasonic pulse group, and the ultrasonic pulse group is transmitted alternately so that the ultrasonic pulse group is transmitted in the other transmission band pair while the ultrasonic pulse group is not transmitted in one transmission band pair. An ultrasonic diagnostic apparatus, characterized in that every time K ultrasonic pulses are transmitted in each transmission band, the machine learning model generates Doppler measurement information for one frame for each transmission band.

6. The ultrasonic diagnostic apparatus according to Claim 5, characterized in that an operation of alternately transmitting ultrasonic pulse groups for two adjacent transmission band pairs is performed a plurality of times while the Doppler measurement information for one frame is being generated.

Images