Ultrasonic image processing device

Abstract

To provide an ultrasonic image processing device capable of automatically setting a proper gamma curve representing a relation between power of a Doppler signal or a velocity represented by the Doppler signal and a pixel value in a Doppler image.SOLUTION: A Doppler signal processing part 30 acquires target power distribution information representing power distribution of Doppler signals. A texture parameter specification part 32 inputs the target power distribution information to a learning model 34. The trained learning model 34 outputs a texture parameter suitable for the target power distribution information. The texture parameter represents a feature of a gamma curve. A proper gamma curve determination part 40 determines a proper gamma curve suitable for the target power distribution information based on the texture parameter suitable for the target power distribution information. An image formation part 18 forms a power Doppler image based on the object power distribution information and the proper gamma curve.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] This specification discloses an improvement to an ultrasound imaging device. [Background technology]

[0002] Conventionally, there is known an ultrasonic diagnostic device that transmits and receives ultrasonic waves toward a subject, forms an ultrasonic tomographic image based on the received signal obtained by the transmission, and displays the formed ultrasonic tomographic image (i.e., B-mode image) on a display. In such an ultrasonic diagnostic device, a Doppler signal is formed from the received signal by utilizing the Doppler effect in the received wave from the subject, and a Doppler image is formed based on the Doppler signal. Doppler images include color Doppler images in which the speed and direction of blood flow in the subject are shown by color, and power Doppler images in which the power of the Doppler signal is expressed by brightness or color. Doppler images are generally displayed superimposed on ultrasonic tomographic images.

[0003] Here, the Doppler signal may contain signal components that represent the movement of the subject's tissue (e.g., the heart wall, etc.) in addition to the blood flow signal that represents the blood flow. Such signal components are called clutter signals. In general, clutter signals become noise components that reduce the visibility of blood flow signals in a Doppler image. Therefore, techniques for suppressing clutter signals have been proposed in the past.

[0004] For example, Patent Document 1 discloses an ultrasound imaging device that discriminates between clutter signals and blood flow signals in an ultrasound reception signal based on a feature that is a parameter (e.g., the amplitude in the packet direction of an IQ signal) that represents the physical difference between the clutter signals and blood flow signals, generates a suppression map that suppresses the clutter signals based on the discrimination result, and applies the suppression map to the reception signal to suppress the clutter signals. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Public Relations, Patent Publication No. 2023-003852 Summary of the Invention [Problem to be solved by the invention]

[0006] As described above, the Doppler image includes an image expressing the power of the Doppler signal by color or brightness (power Doppler image), or an image expressing the blood flow or tissue velocity of the subject indicated by the Doppler signal by color (color Doppler image). Therefore, when forming a Doppler image, the power of the Doppler signal or the velocity indicated by the Doppler signal is converted into a pixel value using a gamma curve indicating the relationship between the power of the Doppler signal or the velocity indicated by the Doppler signal and the pixel value in the Doppler image.

[0007] Fig. 8 is a diagram showing a first example of a gamma curve. In Fig. 8, the horizontal axis represents the power value of the Doppler signal, and the vertical axis represents the frequency or pixel value. The dashed-dotted line graph represents the power value and frequency of the clutter signal contained in the Doppler signal, and the thin solid line graph represents the power value and frequency of the blood flow signal contained in the Doppler signal. In general, the power value of the clutter signal is small and the power value of the blood flow signal is large, so the clutter signals and blood flow signals are distributed as shown in Fig. 8.

[0008] The thick solid line in Fig. 8 is a gamma curve showing the relationship between the power value and the pixel value. According to the gamma curve shown in Fig. 8, a power value smaller than the clutter suppression cutoff has a minimum pixel value PV min Any power value greater than the blood flow signal cutoff is converted to the maximum pixel value PV max That is, the difference in power value in the power region smaller than the clutter suppression cutoff is not expressed by the pixel value, and the difference in power value in the power region larger than the blood flow signal cutoff is not expressed by the pixel value.

[0009] On the other hand, according to the gamma curve in FIG. 8, the power value included in the dynamic range between the clutter suppression cutoff and the blood flow signal cutoff is converted into a pixel value corresponding to that power value. In other words, for the power value between the clutter suppression cutoff and the blood flow signal cutoff, the difference in power value is expressed by the pixel value. This is sometimes referred to as the power value being expressed in gradations. In the example of FIG. 8, the gamma curve within the dynamic range is at point A (power value, pixel value)=(clutter suppression cutoff, PV min ) and point B (power value, pixel value) = (blood flow signal cutoff, PV max ) is a straight line connecting point A and point B, but it may also be a curve connecting point A and point B.

[0010] The gamma curve can be set by the user. In particular, the clutter suppression cutoff and the signal suppression cutoff can be set by the user. Here, unless an appropriate gamma curve is set according to the Doppler signal (more specifically, the power distribution of the clutter signal and the blood flow signal), an appropriate Doppler image (in this case, a power Doppler image) cannot be obtained.

[0011] For example, as shown in Figure 8, if the gamma curve is set so that most of the blood flow signal is greater than the blood flow signal cutoff, most of the blood flow signal will be at the maximum pixel value PV max Therefore, the power value of the blood flow signal cannot be expressed well in gradation in the power Doppler image. Also, as shown in Fig. 9, if the clutter suppression cutoff is set so that many clutter signals are included in the dynamic range, the clutter signals will appear prominently in the power Doppler image, and the visibility of the blood flow signal will decrease.

[0012] In Figures 8 and 9, the horizontal axis represents power value, and gamma curves are shown when a power Doppler image is formed as the Doppler image. However, even when the horizontal axis in Figures 8 and 9 represents the velocity indicated by the Doppler signal (in other words, the Doppler frequency), and a color Doppler image is formed as the Doppler image, the same problem as described above can occur.

[0013] An object of the ultrasonic image processing device disclosed in this specification is to enable automatic setting of an appropriate gamma curve that indicates the relationship between the power of a Doppler signal or the velocity indicated by a Doppler signal and the pixel value in a Doppler image. [Means for solving the problem]

[0014] The ultrasonic image processing device disclosed in the present specification is characterized in that it includes: a texture parameter specifying unit that specifies the texture parameter suitable for target power distribution information by inputting target power distribution information to a learning model that is trained to predict and output the texture parameter suitable for power distribution information when power distribution information is input using, as learning data, a combination of power distribution information representing a power distribution of a Doppler signal obtained by transmitting and receiving ultrasonic waves to a subject, and texture parameters representing a gamma curve suitable for the power distribution information, which indicates a relationship between the power of the Doppler signal and a pixel value in a Doppler image; an appropriate gamma curve determination unit that determines an appropriate gamma curve, which is the gamma curve suitable for the target power distribution information, based on the specified texture parameter; and a Doppler image formation unit that forms a Doppler image in which the power of a Doppler signal is indicated based on the target power distribution information and the appropriate gamma curve.

[0015] The image processing device may further include a learning processing unit that uses a combination of the power distribution information and the texture parameters representing the gamma curve set by a user as learning data to train the learning model so as to predict and output the texture parameters suitable for the user when power distribution information is input, wherein the texture parameter identification unit inputs the target power distribution information to the learning model to identify the texture parameters suitable for the user, and the appropriate gamma curve determination unit determines the appropriate gamma curve suitable for the user based on the identified texture parameters.

[0016] The learning processing unit may present to the user a plurality of combinations of the power distribution information and the texture parameters that are pre-stored in a memory, and learn the learning model by using the combination of the power distribution information and the texture parameters selected by the user as learning data.

[0017] The appropriate gamma curve determination unit may select the appropriate gamma curve from a plurality of types of gamma curves pre-stored in a memory based on the gamma curve calculated on the basis of the texture parameter specified by the texture parameter specification unit.

[0018] The appropriate gamma curve determination unit may notify a user of the determined appropriate gamma curve.

[0019] The ultrasonic image processing device disclosed in the present specification is characterized in that it includes: a texture parameter specifying unit that specifies the texture parameter suitable for target velocity distribution information by inputting target velocity distribution information to a learning model that is trained to predict and output the texture parameter suitable for velocity distribution information when velocity distribution information is input using a combination of velocity distribution information indicating a velocity distribution indicated by a Doppler signal obtained by transmitting and receiving ultrasonic waves to and from a subject, and texture parameters indicating a gamma curve indicating a relationship between the velocity indicated by the Doppler signal and a pixel value in a Doppler image as learning data; an appropriate gamma curve determination unit that determines an appropriate gamma curve, which is the gamma curve suitable for the target velocity distribution information, based on the specified texture parameter; and a Doppler image formation unit that forms a Doppler image indicating the velocity of blood flow or tissue of the subject based on the target velocity distribution information and the appropriate gamma curve. Effect of the Invention

[0020] According to the ultrasound image processing device disclosed in this specification, it is possible to automatically set an appropriate gamma curve that indicates the relationship between the power of a Doppler signal or the velocity indicated by a Doppler signal and the pixel value in a Doppler image. [Brief description of the drawings]

[0021] [Figure 1] 1 is a schematic diagram illustrating the configuration of an ultrasonic diagnostic apparatus according to an embodiment of the present invention. [Diagram 2] FIG. 11 is a diagram showing power distribution information and a gamma curve. [Diagram 3] FIG. 2 is a conceptual diagram showing a learning process of a learning model. [Figure 4] 13 is a flowchart showing the flow of a learning process of a learning model. [Diagram 5] FIG. 13 is a diagram illustrating an example of a learning data selection screen. [Figure 6] FIG. 1 is a conceptual diagram showing the processing of a learning model. [Figure 7] 10 is a flowchart showing the flow of a method for determining an appropriate gamma curve. [Figure 8] FIG. 2 is a second diagram showing an example of a gamma curve. [Figure 9] FIG. 3 is a third diagram showing an example of a gamma curve. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0022] 1 is a schematic diagram of the configuration of an ultrasonic diagnostic apparatus 10 as an ultrasonic image processing apparatus according to this embodiment. The ultrasonic diagnostic apparatus 10 is a medical device installed in a medical institution such as a hospital and used during ultrasonic examinations.

[0023] The ultrasonic diagnostic device 10 is a device that scans an ultrasonic beam on a subject and generates an ultrasonic image based on the received signal obtained by scanning the subject. In particular, the ultrasonic diagnostic device 10 can form an ultrasonic tomographic image (B-mode image) in which the amplitude intensity of the reflected wave from the scanning surface is converted into brightness based on the received signal, and a Doppler image formed based on the difference in frequency between the transmitted wave and the received wave (Doppler shift). In this specification, the Doppler image is a concept that includes a color Doppler image in which the blood flow or the movement speed of the tissue of the subject is shown by color, and a power Doppler image in which the intensity of the Doppler signal is represented by color or brightness.

[0024] The ultrasonic probe 12 is a device that transmits and receives ultrasonic waves to and from a subject. The ultrasonic probe 12 has a transducer array that includes a plurality of transducer elements that transmit and receive ultrasonic waves to and from the subject.

[0025] The transmitting / receiving unit 14 transmits a transmission signal to the ultrasonic probe 12 (specifically, each transducer element of the transducer element array) under the control of a control unit 28 (described later). As a result, ultrasonic waves are transmitted from each transducer element to the subject.

[0026] The transmitting / receiving unit 14 also receives a reception signal from each transducer element that has received a reflected wave from the subject. The transmitting / receiving unit 14 has an adder and a plurality of delays corresponding to each transducer element, and performs a phasing addition process in which the adder and the plurality of delays align and add the phases of the reception signals from each transducer element. This forms a reception beam signal in which information indicating the signal strength and frequency of the reflected wave from the subject is aligned in the depth direction of the subject.

[0027] The signal processing unit 16 performs various signal processing on the received beam signal from the transmitting / receiving unit 14, including filtering using a bandpass filter and detection processing.

[0028] The image forming unit 18 forms an ultrasonic tomographic image (B-mode image) based on the received beam signal that has been signal-processed in the signal processing unit 16. The image forming unit 18 also forms a Doppler image based on a Doppler signal obtained by a Doppler signal processing unit 30 (described later) and a gamma curve determined by an appropriate gamma curve determining unit 40 (described later).

[0029] The display control unit 20 controls the display of the ultrasonic tomographic image and Doppler image formed by the image forming unit 18, as well as various other information, on the display 22. The display 22 as a display unit is a display device constituted by, for example, a liquid crystal display or an organic EL (Electro Luminescence) display.

[0030] The input interface 24 is configured with, for example, a button, a trackball, a touch panel, etc. The input interface 24 is used to input a user's command to the ultrasound diagnostic apparatus 10.

[0031] The memory 26 includes a hard disk drive, a solid state drive (SSD), an embedded multi media card (eMMC), a read only memory (ROM), etc. An ultrasound diagnostic program for operating each unit of the ultrasound diagnostic device 10 is stored in the memory 26. The ultrasound diagnostic program can also be stored in a computer-readable non-transitory storage medium such as a universal serial bus (USB) memory or a CD-ROM. The ultrasound diagnostic device 10 can read and execute the ultrasound diagnostic program from such a storage medium.

[0032] 1, the memory 26 also stores a learning model 34, a learning data candidate DB (DataBase) 38, and a gamma curve DB 42. These will be described in detail later.

[0033] The control unit 28 includes at least one of a general-purpose processor (e.g., a CPU (Central Processing Unit)) and a dedicated processor (e.g., a GPU (Graphics Processing Unit), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a programmable logic device). The control unit 28 may not be a single processing device, but may be configured by the cooperation of multiple processing devices located at physically separate locations. The control unit 28 controls each part of the ultrasound diagnostic apparatus 10 according to an ultrasound diagnostic program stored in the memory 26.

[0034] The Doppler signal processor 30 forms a Doppler signal indicating a Doppler shift (the difference between the transmission frequency of the ultrasound transmitted to the subject and the frequency of the received beam signal) by known Doppler processing. The Doppler signal processor 30 also calculates the power of the Doppler signal by integrating the Doppler signal in the frequency direction. Furthermore, the Doppler signal processor 30 applies a wall filter (low-cut filter) to reduce low-frequency signal components such as the subject's body movement contained in the Doppler signal. In this embodiment, the Doppler signal processor 30 calculates the power of the Doppler signal, and the texture parameter specification unit 32 and the appropriate gamma curve determination unit 40 described later perform processing based on the power distribution of the Doppler signal.

[0035] The texture parameter specifying unit 32 specifies texture parameters suitable for power distribution information indicating the power distribution of the Doppler signal, based on the power distribution information.

[0036] The power distribution of the Doppler signal may be, for example, a graph of the clutter signal CS and the blood flow signal BF in a two-dimensional space in which the horizontal axis indicates the power value and the vertical axis indicates the frequency, as shown in Fig. 2. The power distribution of the Doppler signal may be any information that indicates the frequency relative to the power value. For example, the power distribution of the Doppler signal may be blood flow power distribution information obtained by filtering the Doppler signal to suppress clutter components. The power distribution of the Doppler signal may be the power Doppler image itself.

[0037] The texture parameter is a parameter indicating the characteristics of the gamma curve GC (see FIG. 2). As described above, the gamma curve GC is information indicating the relationship between a Doppler signal (power value in this embodiment) and a pixel value in a Doppler image (power Doppler image in this embodiment). In this embodiment, the texture parameter includes a parameter indicating the ratio of the clutter signal CS (integral value of the clutter signal CS within the dynamic range D in FIG. 2) included in the dynamic range D defined by the gamma curve GC to the blood flow signal BF (integral value of the entire blood flow signal BF in FIG. 2), and a parameter indicating the range of the blood flow signal BF to be included in the dynamic range D (in other words, blood flow signal cutoff BFC).

[0038] The texture parameter specifying unit 32 specifies the texture parameters by using the learning model 34. Here, the learning model 34 and a learning method of the learning model 34 by the learning processing unit 36 ​​will be described.

[0039] The learning model 34 is composed of, for example, a CNN (Convolutional Neural Network). The learning model 34 is not limited to a CNN, and may be any model as long as it exhibits the functions described below. The learning model 34 uses a combination of power distribution information and texture parameters representing a gamma curve suitable for the power distribution information as learning data, and is trained to predict and output texture parameters suitable for the power distribution information when the power distribution information is input.

[0040] The learning model 34 may be trained by a device other than the ultrasound diagnostic device 10, and the trained learning model 34 may be stored in the memory 26. However, it is conceivable that the appropriate (in other words, preferred) gamma curve differs for each facility (e.g., hospital) in which the ultrasound diagnostic device 10 is installed. Therefore, the ultrasound diagnostic device 10 according to this embodiment has a learning processing unit 36, and the learning processing unit 36 ​​executes the learning process of the learning model 34. As a result, it becomes possible to output texture parameters suitable for each ultrasound diagnostic device 10 using a different trained learning model 34 for each ultrasound diagnostic device 10. In addition, in order to reduce the load of the learning process in each ultrasound diagnostic device 10 or the amount of learning data for learning, the learning model 34 may be trained in advance by a device other than the ultrasound diagnostic device 10, and the learning processing unit 36 ​​may perform fine tuning by re-training the learning model 34.

[0041] The learning data used for learning the learning model 34 is a combination of power distribution information and texture parameters that represent a gamma curve appropriate for the power distribution information (for example, manually set by a user in the past). As shown in FIG. 3, the learning processing unit 36 ​​inputs the power distribution information included in the learning data to the learning model 34. The learning model 34 predicts and outputs texture parameters appropriate for the input power distribution information. Based on the difference between the texture parameters output by the learning model 34 and the texture parameters included in the learning data, the learning processing unit 36 ​​adjusts the parameters of the learning model 34 so that the difference becomes smaller. By repeating such processing, the learning model 34 is trained. The learning model 34 that has been sufficiently trained becomes able to output texture parameters appropriate for the power distribution information with high accuracy based on the input power distribution information.

[0042] As described above, the learning processing unit 36 ​​of the ultrasound diagnostic device 10 trains the learning model 34, so that the trained learning model 34 can output texture parameters suitable for the ultrasound diagnostic device 10 (in other words, a certain facility), but when multiple users (e.g., doctors, technicians, etc.) use the same ultrasound diagnostic device 10, it is possible that the texture parameters suitable for each user are different (in other words, each user has a different preference for a gamma curve). Therefore, it is preferable that the learning model 34 is trained to predict and output texture parameters suitable for the user who uses the ultrasound diagnostic device 10.

[0043] Specifically, the learning processing unit 36 ​​uses a combination of power distribution information and texture parameters representing a gamma curve set by the user of the ultrasound diagnostic device 10 as learning data. Furthermore, the learning data may include user information (e.g., a user ID) indicating the user. In this case, the learning processing unit 36 ​​inputs the power distribution information and the user information to the learning model 34. The learning model 34 predicts and outputs texture parameters suitable for the user indicated by the input power distribution information and the input user information. Based on a difference between the texture parameters output by the learning model 34 and the texture parameters included in the learning data representing the gamma curve set by the user, the learning processing unit 36 ​​adjusts the parameters of the learning model 34 so that the difference is reduced. The learning model 34 thus learned can output texture parameters suitable for the user indicated by the power distribution information and the user information with high accuracy based on the input power distribution information and user information.

[0044] Alternatively, a different learning model 34 may be prepared for each user. In this case, a plurality of learning models 34 corresponding to a plurality of users are stored in the memory 26. The learning processing unit 36 ​​uses a combination of the power distribution information and a texture parameter representing a gamma curve set by the user of the ultrasound diagnostic device 10 as learning data for learning the learning model 34 corresponding to the user.

[0045] There may be cases where it is difficult to obtain a sufficient amount of learning data for the learning processing unit 36 ​​to learn the learning model 34, or where it takes a considerable amount of time to obtain a sufficient amount of learning data. Therefore, it is preferable to prepare in advance a plurality of combinations of power distribution information and texture parameters (hereinafter, referred to as "learning data candidates") that can be used as learning data, and allow the user to select desired learning data from the plurality of learning data candidates. Hereinafter, the flow of the learning process of the learning model 34 using learning data selected from the learning data candidates will be described according to the flowchart shown in FIG. 4.

[0046] In this embodiment, the multiple learning data candidates are stored in advance in the learning data candidate DB 38. In step S10, the multiple learning data candidates are presented to the user. In this embodiment, in response to an instruction from the user, as shown in FIG. 5, the display control unit 20 displays a learning data selection screen on the display 22 on which multiple learning data candidates are displayed. In the example of FIG. 5, a power Doppler image is displayed as power distribution information of the learning data candidates, and texture parameters suitable for the power distribution information are displayed below (associated with) the power Doppler image. In the example of FIG. 5, two learning data candidates are displayed, but the display control unit 20 may of course display more than one learning data candidate on the display 22. If the multiple learning data candidates cannot be displayed on one screen, the display control unit 20 can display multiple learning data candidates by switching the screen in response to a user operation.

[0047] In step S12, the user selects desired learning data from the learning data selection screen. At this time, the learning data selected by the user can be regarded as learning data including texture parameters suitable for the user.

[0048] In step S14, the learning processing unit 36 ​​sets, as learning data, a combination of a power Doppler image serving as power distribution information selected by the user on the learning data selection screen and a texture parameter associated with the power Doppler image.

[0049] In step S16, the learning processing unit 36 ​​uses the learning data identified in step S14 to train the learning model 34. When the learning model 34 is provided for each user, after step S16, a process of associating the learning model 34 with the user is performed.

[0050] 4 may be repeatedly executed. As a result, the learning model 34 that has been sufficiently trained can output texture parameters suitable for the user indicated by the input power distribution information and user information with high accuracy, based on the input power distribution information and user information.

[0051] In the examples of Figs. 4 and 5, the learning data candidates include power Doppler images as power distribution information, but the power distribution information included in the learning data candidates may be the graphs of the clutter signal CS and the blood flow signal BF shown in Fig. 2 or the above-mentioned blood flow power distribution information.

[0052] As shown in FIG. 6, the texture parameter specifying unit 32 inputs power distribution information of the processing target to a learning model 34 that has been sufficiently trained by the above-mentioned learning process. In this specification, the power distribution information of the processing target is called the target power distribution information. The target power distribution information may also be a graph of the clutter signal CS and the blood flow signal BF shown in FIG. 2, the above-mentioned blood flow power distribution information, or a Doppler power image. The trained learning model 34 outputs texture parameters suitable for the target power distribution information. The texture parameter specifying unit 32 specifies texture parameters suitable for the target power distribution information based on the output of the learning model 34.

[0053] As described above, when the learning model 34 is trained using training data including power distribution information, user information, and texture parameters, the texture parameter specification unit 32 may input the target power distribution information and user information indicating the user currently using the ultrasound diagnostic device 10 to the trained learning model 34. The user information can be acquired by the control unit 28 authenticating the user (for example, by inputting a user ID and password). This allows the texture parameter specification unit 32 to specify texture parameters suitable for the user based on the output of the learning model 34.

[0054] Furthermore, when a different learning model 34 is prepared for each user, the texture parameter specifying unit 32 specifies the learning model 34 to which the target power distribution information is to be input, based on user information indicating the user currently using the ultrasound diagnostic device 10. Then, the texture parameter specifying unit 32 inputs the target power distribution information to the specified learning model 34, and specifies texture parameters suitable for the user based on the output of the learning model 34.

[0055] The appropriate gamma curve determination unit 40 determines a gamma curve appropriate for the target power distribution information based on the texture parameters appropriate for the target power distribution information identified by the texture parameter identification unit 32. In this specification, a gamma curve appropriate for the target power distribution information is referred to as an appropriate gamma curve. As described above, the texture parameters are parameters that indicate the characteristics of a gamma curve, so the appropriate gamma curve determination unit 40 can determine an appropriate gamma curve based on the texture parameters appropriate for the target power distribution information.

[0056] As described above, in this embodiment, the texture parameters include a parameter indicating the ratio of the clutter signal CS (the integral value of the clutter signal CS within the dynamic range D in FIG. 2) included in the dynamic range D defined by the gamma curve GC to the blood flow signal BF (the integral value of the entire blood flow signal BF in FIG. 2), and a parameter indicating the range of the blood flow signal BF to be included in the dynamic range D (in other words, the blood flow signal cutoff BFC), so a method for determining an appropriate gamma curve from the texture parameters will be described.

[0057] 2, the gamma curve GC is determined by a blood flow signal cutoff BFC and a clutter signal cutoff CSC. Therefore, the appropriate gamma curve determiner 40 determines the appropriate gamma curve by determining the blood flow signal cutoff BFC and the clutter signal cutoff CSC based on the texture parameters.

[0058] First, regarding the blood flow signal cutoff BFC, since the texture parameters include the blood flow signal cutoff BFC, the appropriate gamma curve determination unit 40 can directly determine the blood flow signal cutoff BFC from the texture parameters.

[0059] The texture parameter also includes the proportion of the clutter signals CS included within the dynamic range D to the entire blood flow signal BF in the target power distribution information. The clutter signals CS included within the dynamic range D are a portion of the entire clutter signals CS having a power value greater than the clutter signal cutoff CSC. Therefore, the appropriate gamma curve determiner 40 determines the clutter signal cutoff CSC so that the proportion of the portion of the entire clutter signals CS having a power value greater than the clutter signal cutoff CSC to the entire blood flow signal BF in the target power distribution information is the proportion indicated by the texture parameter.

[0060] The appropriate gamma curve determination unit 40 determines the pixel value for the power value smaller than the clutter signal cutoff CSC as a minimum pixel value PV min(For example, the brightness value is 0), and the pixel value for the power value larger than the blood flow signal cutoff BFC is the maximum pixel value PV max (for example, maximum brightness), and a proper gamma curve is determined so that pixel values ​​for power values ​​between the clutter signal cutoff CSC and the blood flow signal cutoff BFC, in other words, power values ​​included in the dynamic range D, correspond to the power values. In this embodiment, the proper gamma curve within the dynamic range D is point A (power value, pixel value) = (clutter suppression cutoff, PV min ) and point B (power value, pixel value) = (blood flow signal cutoff, PV max ) is a straight line connecting

[0061] As described above, when the learning model 34 is trained to predict and output texture parameters suitable for a user, the appropriate gamma curve determination unit 40 can determine an appropriate gamma curve suitable for the user based on the texture parameters identified by the texture parameter identification unit 32.

[0062] In the memory 26, a gamma curve DB 42 in which a plurality of types of gamma curves are registered in advance may be stored. In such an ultrasonic diagnostic device 10, only gamma curves registered in the gamma curve DB 42 may be settable. In this case, the appropriate gamma curve determined by the appropriate gamma curve determination unit 40 may not be registered in the gamma curve DB 42. Therefore, in this case, the appropriate gamma curve determination unit 40 may select an appropriate gamma curve from the plurality of types of gamma curves stored in the gamma curve DB 42 based on a gamma curve calculated based on the texture parameters specified by the texture parameter specification unit 32. For example, the appropriate gamma curve determination unit 40 may determine, as the appropriate gamma curve, a gamma curve that is most similar to the gamma curve obtained by the appropriate gamma curve determination unit 40 based on the texture parameters among the plurality of types of gamma curves registered in the gamma curve DB 42. The similarity of the gamma curves may be calculated, for example, according to the difference in clutter signal cutoff CSC and the difference in blood flow signal cutoff BFC between both gamma curves.

[0063] The image forming unit 18 as a Doppler image forming unit forms a power Doppler image based on the target power distribution information and the appropriate gamma curve determined by the appropriate gamma curve determining unit 40. The power Doppler image formed in this manner can be an image in which the blood flow signal BF is appropriately expressed in gradations and the influence of the clutter signal CS is reduced. The appropriate gamma curve determining unit 40 may notify the user of the determined appropriate gamma curve. Notifying the user of the appropriate gamma curve may mean presenting a graph of the gamma curve GC as shown in FIG. 2 to the user, or presenting the user with a Doppler image formed based on the determined appropriate gamma curve.

[0064] The appropriate gamma curve determination unit 40 may notify the user of the determined appropriate gamma curve. For example, the appropriate gamma curve determination unit 40 may present the user with a graph of the gamma curve GC as shown in FIG. 2 in the form of a pop-up. In particular, the appropriate gamma curve determination unit 40 may notify the user of the determined appropriate gamma curve when a gamma curve previously determined in a similar scene (for example, when a similar cross section is displayed or when the similar examination content is performed) is different from the determined appropriate gamma curve. In this case, the image forming unit 18 may form a Doppler image upon receiving an instruction from the user who has been notified of the appropriate gamma curve.

[0065] In the above embodiment, the processing for forming a power Doppler image as the Doppler image has been described, but the present invention can also be applied to the case where a color Doppler image is formed as the Doppler image.

[0066] In this case, the gamma curve indicates the relationship between the velocity indicated by the Doppler signal and the pixel value in the color Doppler image. The learning model 34 uses a combination of velocity distribution information indicating the velocity distribution indicated by the Doppler signal (which may be called the frequency distribution of the Doppler signal) and a texture parameter indicating the gamma curve as learning data, and is trained to predict and output texture parameters suitable for the velocity distribution information when the velocity distribution information is input.

[0067] The texture parameter specifying unit 32 inputs target speed distribution information, which is speed distribution information of a processing target, to a trained learning model 34, thereby specifying texture parameters suitable for the target speed distribution information.

[0068] The appropriate gamma curve determination unit 40 determines an appropriate gamma curve suitable for the target speed distribution information, based on the texture parameters identified by the texture parameter identification unit 32. In this case, the method of determining the appropriate gamma curve may be the same as that described above.

[0069] Then, the image forming unit 18 as a Doppler image forming unit forms a color Doppler image showing the velocity of the blood flow or tissue of the subject based on the target velocity distribution information and the determined appropriate gamma curve.

[0070] In addition, in the above-described embodiment, the pixel values ​​of the entire Doppler image are optimized based on an appropriate gamma curve that is suitable for power distribution information indicating the power distribution of the entire Doppler signal, or velocity distribution information indicating the velocity distribution indicated by the entire Doppler signal, but it is also possible to optimize the pixel values ​​of a portion of the Doppler image.

[0071] In this case, the image forming unit 18 forms a Doppler image before the gamma curve adjustment, and the display control unit 20 displays the Doppler image on the display 22. The user specifies a portion of the Doppler image that is to be optimized, for example, using a cursor. The portion is called the optimized portion. Then, the texture parameter specifying unit 32 inputs only a portion of the Doppler signal that corresponds to the optimized portion to the learning model 34, and specifies texture parameters based on the output of the learning model 34. The appropriate gamma curve determining unit 40 determines a partial appropriate gamma curve, which is an appropriate gamma curve for the optimized portion, based on the specified texture parameters. The image forming unit 18 forms a Doppler image for the optimized portion using the determined partial appropriate gamma curve. For portions other than the optimized portion, a predetermined gamma curve (for example, specified by the user) may be used, or an appropriate gamma curve determined based on power distribution information or velocity distribution information related to the entire Doppler signal may be used as in the above-mentioned embodiment.

[0072] Furthermore, the optimization part may be determined automatically by the ultrasound diagnostic device 10 without being instructed by a user. For example, the ultrasound tomographic image (B-mode image) formed by the image forming unit 18 may be analyzed using a known technique to automatically detect the blood vessel position, and a predetermined region including the blood vessel position may be automatically set as the optimization part.

[0073] The configuration of the ultrasound diagnostic device 10 according to this embodiment is as described above. The transmission / reception unit 14, the signal processing unit 16, the image forming unit 18, the display control unit 20, the Doppler signal processing unit 30, the texture parameter identification unit 32, the learning processing unit 36, and the appropriate gamma curve determination unit 40 of the ultrasound diagnostic device 10 are configured by a processor. The processor includes at least one of a general-purpose processing unit (e.g., a CPU (Central Processing Unit)) and a dedicated processing unit (e.g., a GPU (Graphics Processing Unit), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a programmable logic device). The processor may not be a single processing unit, but may be configured by the cooperation of multiple processing units located at physically separate positions. Furthermore, each of the above units may be realized by the cooperation of hardware such as a processor and software.

[0074] Hereinafter, the flow of the process of determining an appropriate gamma curve will be described with reference to the flowchart shown in Fig. 7. At the start of the flowchart shown in Fig. 7, it is assumed that the learning model 34 has been sufficiently learned.

[0075] In step S20, the Doppler signal processor 30 performs Doppler processing on the received beam data from the transmitter / receiver 14 to form a Doppler signal to be processed. Here, the Doppler signal processor 30 calculates the power of the Doppler signal. As a result, target power distribution information is obtained.

[0076] In step S22, the texture parameter specifying unit 32 inputs the target power distribution information acquired in step S10 to the learning model .

[0077] In step S24, the learning model 34 predicts and outputs texture parameters suitable for the input target power distribution information. The texture parameter specification unit 32 specifies texture parameters suitable for the target power distribution information based on the output of the learning model 34.

[0078] In step S26, the appropriate gamma curve determination unit 40 determines an appropriate gamma curve suitable for the target power distribution information, based on the texture parameters identified in step S24.

[0079] In step S28, the image forming unit 18 forms a power Doppler image based on the target power distribution information and the appropriate gamma curve determined in step S26.

[0080] Although the embodiment of the present invention has been described above, the present invention is not limited to the above embodiment, and various modifications are possible without departing from the spirit of the present invention.

[0081] For example, in the above embodiment, the ultrasonic diagnostic device 10 is an ultrasonic image processing device, but the ultrasonic image processing device is not limited to the ultrasonic diagnostic device 10. For example, the ultrasonic image processing device may be a personal computer or the like. In this case, the ultrasonic image processing device is provided with the functions of an image forming unit 18, a display control unit 20, a display 22, an input interface 24, a memory 26, a control unit 28, a Doppler signal processing unit 30, a texture parameter specifying unit 32, a learning processing unit 36, and an appropriate gamma curve determining unit 40. The ultrasonic image processing device receives reception beam data from the ultrasonic diagnostic device, and the above units can execute processing. [Explanation of symbols]

[0082] 10 ultrasound diagnostic device, 12 ultrasound probe, 14 transmitter / receiver unit, 16 signal processing unit, 18 image formation unit, 20 display control unit, 22 display, 24 input interface, 26 memory, 28 control unit, 30 Doppler signal processing unit, 32 texture parameter identification unit, 34 learning model, 36 learning processing unit, 38 learning data candidate DB, 40 appropriate gamma curve determination unit, 42 gamma curve DB.

Claims

1. a texture parameter specifying unit that uses, as learning data, a combination of power distribution information representing a power distribution of a Doppler signal obtained by transmitting and receiving ultrasound to a subject, and a texture parameter representing a gamma curve suitable for the power distribution information, which indicates a relationship between the power of the Doppler signal and a pixel value in a Doppler image, and that specifies the texture parameter suitable for the target power distribution information by inputting target power distribution information that is a processing target, into a learning model that has been trained to predict and output the texture parameter suitable for the power distribution information when power distribution information is input; an appropriate gamma curve determination unit that determines an appropriate gamma curve, which is the gamma curve that is appropriate for the target power distribution information, based on the specified texture parameter; a Doppler image forming unit that forms a Doppler image showing the power of a Doppler signal based on the target power distribution information and the appropriate gamma curve; 13. An ultrasonic image processing device comprising:

2. a learning processing unit that uses a combination of the power distribution information and the texture parameter representing the gamma curve set by a user as learning data, and trains the learning model so as to predict and output the texture parameter suitable for the user when power distribution information is input; Further equipped with the texture parameter specification unit specifies the texture parameter suitable for the user by inputting the target power distribution information to the learning model; the appropriate gamma curve determination unit determines the appropriate gamma curve suitable for the user based on the identified texture parameter.

2. The ultrasonic image processing apparatus according to claim 1.

3. the learning processing unit presents to the user a plurality of combinations of the power distribution information and the texture parameter, which are stored in advance in a memory, and learns the learning model by using the combination of the power distribution information and the texture parameter selected by the user as learning data.

3. The ultrasonic image processing apparatus according to claim 2.

4. the appropriate gamma curve determination unit selects the appropriate gamma curve based on the gamma curve calculated based on the texture parameter specified by the texture parameter specification unit, from among a plurality of types of gamma curves stored in advance in a memory.

2. The ultrasonic image processing apparatus according to claim 1.

5. the appropriate gamma curve determination unit notifies a user of the determined appropriate gamma curve. do, 2. The ultrasonic image processing apparatus according to claim 1.

6. a texture parameter specifying unit that uses, as learning data, a combination of velocity distribution information indicating a velocity distribution indicated by a Doppler signal obtained by transmitting and receiving ultrasonic waves to a subject, and a texture parameter indicating a gamma curve indicating a relationship between the velocity indicated by the Doppler signal and a pixel value in a Doppler image, and that specifies the texture parameter appropriate for the target velocity distribution information by inputting target velocity distribution information that is a processing target into a learning model that has been trained to predict and output the texture parameter appropriate for the velocity distribution information when the velocity distribution information is input; an appropriate gamma curve determination unit that determines an appropriate gamma curve, which is the gamma curve that is appropriate for the target speed distribution information, based on the specified material appearance parameter; a Doppler image forming unit that forms a Doppler image showing the velocity of blood flow or tissue of the subject based on the target velocity distribution information and the appropriate gamma curve; 13. An ultrasonic image processing device comprising:

Images