JP7309355B2 - Medical image generation device and medical image generation program - Google Patents

Description

TECHNICAL FIELD Embodiments of the present invention relate to a medical image generation apparatus and a medical image generation program.

Conventionally, an ultrasonic diagnostic apparatus has a color mode that expresses blood flow in colors. The operator must set appropriate parameters in order to generate image data showing accurate blood flow in color mode.

However, appropriate parameters vary depending on various factors such as the physique of the subject, the depth of the region of the subject to be inspected, the organ to be inspected, etc. Therefore, it is difficult to generate image data with appropriate parameters. there were.

U.S. Patent Application Publication No. 2016/0174943 U.S. Patent Application Publication No. 2018/0028079 U.S. Patent Application Publication No. 2007/0014452

The problem to be solved by the present invention is to generate image data with appropriate parameters.

A medical image generating apparatus according to an embodiment includes an acquisition unit, a setting unit, a processing unit , and a determination unit. The acquisition unit acquires subject image data, which is an image including the subject. The setting unit sets at least two regions of interest having different depths of the subject with respect to the subject image data . The processing unit inputs the object image data to a trained model that has learned the correlation between the image data and the evaluation value for each depth of the image data, and calculates the evaluation value of the object image data . Output information is obtained for each of the regions of interest . The determination unit determines a parameter relating to acquisition of the subject image data so that the evaluation value indicated for each region of interest in the output information obtained by the processing unit satisfies a reference value .

FIG. 1 is a diagram showing an example of the configuration of a medical information processing system according to the first embodiment. FIG. 2 is a block diagram showing an example of the configuration of the medical image generating apparatus according to the first embodiment; FIG. 3 is an explanatory diagram for explaining a region of interest set in ultrasound image data. FIG. 4 is a diagram showing learning processing by the first learning function. FIG. 5 is a diagram showing operation processing by the evaluation processing function. FIG. 6 is a flow chart showing a processing procedure of image generation processing according to the first embodiment. FIG. 7 is a block diagram showing an example of a configuration of a medical image generating apparatus according to a first modification of the first embodiment; FIG. 8 is a block diagram showing an example configuration of a medical image generating apparatus according to the second embodiment. FIG. 9 is a diagram showing learning processing by the second learning function. FIG. 10 is a diagram showing operation processing by the estimation processing function. FIG. 11 is a flow chart showing a processing procedure of image generation processing according to the second embodiment. FIG. 12 is a block diagram showing an example of the configuration of a medical image generating apparatus according to the first modified example of the second embodiment;

Hereinafter, embodiments of a medical image generation apparatus and a medical image generation program will be described with reference to the drawings. Note that the contents described in one embodiment or modified example may be similarly applied to other embodiments or other modified examples.

(First embodiment)
A first embodiment will be described. FIG. 1 is a diagram showing an example of the configuration of a medical information processing system 1 according to the first embodiment. As shown in FIG. 1 , the medical information processing system 1 includes a medical image generation device 10 , an electronic chart system 20 , a PACS (Picture Archiving and Communication System) 30 , and a medical information processing device 40 . Each system and each device are communicably connected via a network 50 . The configuration shown in FIG. 1 is merely an example, and devices other than the illustrated medical image generating apparatus 10, electronic medical record system 20, PACS 30, and medical information processing apparatus 40 may be connected to the network 50.

The medical image generating apparatus 10 images a subject and generates a medical image. More specifically, the medical image generating apparatus 10 images the subject according to the order information transmitted from the electronic medical record system 20 and generates a medical image. Then, the medical image generating apparatus 10 transmits the generated medical image to the PACS 30 each time it generates a medical image. For example, the medical image generating apparatus 10 includes an X-ray diagnostic apparatus, an X-ray CT (Computed Tomography) apparatus, an MRI (Magnetic Resonance Imaging) apparatus, an ultrasonic diagnostic apparatus, a SPECT (Single Photon Emission Computed Tomography) apparatus, a PET (Positron Emission computed tomography) equipment, etc.

The electronic medical record system 20 manages various types of information about subjects. More specifically, the electronic medical record system 20 serves as an ordering system, and receives input of various orders regarding subjects from operators such as doctors and technicians. The electronic medical record system 20 transmits order information indicating the details of the received order to department systems provided in examination rooms, pharmacies, radiology departments, and the like. Then, the electronic medical record system 20 acquires medical care data transmitted from each department system as a result of medical care performed according to the order information. For example, the electronic medical record system 20 transmits order information instructing imaging of medical images to the medical image generating apparatus 10 . The electronic medical record system 20 acquires medical images captured by the medical image generating apparatus 10 according to instructions. For example, the electronic medical record system 20 is implemented by computer equipment such as servers and workstations.

The PACS 30 stores medical images generated by the medical image generating apparatus 10 . More specifically, PACS 30 receives medical images from medical image generation device 10 . The PACS 30 stores and manages the received medical images in its own memory circuit or the like. For example, the PACS 30 is realized by computer equipment such as servers and workstations.

The medical information processing apparatus 40 acquires various kinds of information from the electronic medical record system 20 , the medical image generating apparatus 10 and the PACS 30 . Then, the medical information processing apparatus 40 performs various information processing using the acquired information. For example, the medical information processing apparatus 40 is implemented by computer equipment such as a server, workstation, personal computer, tablet terminal, or the like.

Such a medical information processing system 1 learns the correlation between image data generated by a device such as the medical image generating device 10 and parameters relating to acquisition of the image data by machine learning. Then, the medical image generating apparatus 10 generates image data with appropriate parameters by adjusting the parameters based on the learned model constructed by machine learning.

Here, the parameters are settings related to image data acquisition by the medical image generating apparatus 10 . For example, the parameters are settings related to acquisition of ultrasound image data in a color mode for displaying ultrasound image data expressing blood flow in colors in ultrasound diagnosis. More specifically, the parameters include acquisition parameters set as acquisition conditions before various data are acquired in ultrasonic diagnosis, and post-processing parameters used in post-processing applied to images after various data are acquired. Both may be included. For example, parameters include ultrasound output conditions, gain magnitude, imaging conditions, settings for specifying the range of blood flow velocities to be displayed, and suppression of erroneous detection of small movements of the subject as blood flow. filters and the like.

Machine learning includes deep learning, neural network, logistic regression analysis, nonlinear discriminant analysis, support vector machine (SVM), random forest, naive Naive Bayes et al.

The medical image generating apparatus 10 according to the first embodiment uses a trained model constructed by inputting image data and an evaluation value indicating evaluation of the image data as learning data into a machine learning engine, Output the evaluation value of the generated image data. Further, when the evaluation value of the generated image data is less than the threshold value, the medical image generating apparatus 10 changes the parameters and generates the image data. Then, the medical image generating apparatus 10 generates appropriate image data by repeating the process until the evaluation value of the generated image data becomes equal to or greater than the threshold value, or by repeating the process up to a predetermined number of times. The medical image generating apparatus 10 limits the generation of image data with changed parameters to a predetermined number of times, thereby preventing the processing of generating image data by changing parameters from never ending.

In this embodiment, a case where the medical image generating apparatus 10 is an ultrasonic diagnostic apparatus will be described as an example.

FIG. 2 is a block diagram showing an example of the configuration of the medical image generating apparatus 10 according to the first embodiment. As shown in FIG. 2, the medical image generating apparatus 10 according to the first embodiment has an apparatus main body 100, an ultrasonic probe 101, an input interface 102, and a display 103. The ultrasonic probe 101, the input interface 102, and the display 103 are communicably connected to the apparatus body 100. FIG.

The ultrasonic probe 101 has multiple piezoelectric transducers. These piezoelectric vibrators generate ultrasonic waves based on drive signals supplied from the transmission/reception circuit 110 of the device body 100 . Also, the ultrasonic probe 101 receives reflected waves from the subject P and converts them into electrical signals. That is, the ultrasonic probe 101 scans the subject P with ultrasonic waves and receives reflected waves from the subject P. FIG. Further, the ultrasonic probe 101 has a matching layer provided on the piezoelectric transducers, a backing material, etc. for preventing the ultrasonic waves from propagating backward from the piezoelectric transducers. Note that the ultrasonic probe 101 is detachably connected to the device main body 100 .

When ultrasonic waves are transmitted from the ultrasonic probe 101 to the subject P, the transmitted ultrasonic waves are successively reflected by discontinuous surfaces of acoustic impedance in the body tissue of the subject P, and are reflected as reflected wave signals from the ultrasonic probe. The signal is received by a plurality of piezoelectric vibrators 101 has. The amplitude of the received reflected wave signal depends on the difference in acoustic impedance at the discontinuity from which the ultrasonic waves are reflected. When the transmitted ultrasonic pulse is reflected by the moving blood flow or the surface of the heart wall, the reflected wave signal depends on the velocity component of the moving object in the ultrasonic transmission direction due to the Doppler effect. subject to frequency shifts.

In this embodiment, the ultrasonic probe 101 may be a 1D array probe that scans the subject P two-dimensionally, or a three-dimensional probe that scans the subject P three-dimensionally, that is, a mechanical 4D probe or a 2D array probe. is also applicable.

The input interface 102 is a trackball, switch button, mouse, keyboard, and touch panel for setting predetermined positions (for example, tissue shape position, region of interest, regions other than the region of interest, etc.). It is realized by a touch pad for performing input operations, a touch monitor in which a display screen and a touch pad are integrated, a non-contact input circuit using an optical sensor, a voice input circuit, and the like. The input interface 102 is connected to a processing circuit 160 to be described later, converts an input operation received from an operator (user) into an electric signal, and outputs the electric signal to the processing circuit 160 . It should be noted that the input interface 102 in this specification is not limited to having physical operation parts such as a mouse and a keyboard. For example, the input interface includes an electrical signal processing circuit that receives an electrical signal corresponding to an input operation from an external input device provided separately from the device and outputs the electrical signal to the control circuit.

The display 103 displays a GUI (Graphical User Interface) for an operator (user) of the medical image generating apparatus 10 to input various setting requests using the input interface 102 , and displays ultrasonic waves generated in the apparatus main body 100 . It displays image data, etc. Further, the display 103 displays various messages and display information in order to notify the operator of the processing status and processing results of the apparatus main body 100 . The display 103 also has a speaker and can output sound.

The device main body 100 is a device that generates ultrasonic image data based on reflected wave signals received by the ultrasonic probe 101 . The device main body 100 shown in FIG. 2 is a device capable of generating two-dimensional ultrasonic image data based on two-dimensional reflected wave data (echo data) received by the ultrasonic probe 101 . Further, the apparatus main body 100 shown in FIG. 2 is an apparatus capable of generating three-dimensional ultrasonic image data (volume data) based on three-dimensional reflected wave data received by the ultrasonic probe 101 .

The device body 100 has a transmission/reception circuit 110, a B-mode processing circuit 120, a Doppler processing circuit 130, a storage circuit 140, a communication interface 150, and a processing circuit 160, as shown in FIG. The transmission/reception circuit 110, the B-mode processing circuit 120, the Doppler processing circuit 130, the storage circuit 140, the communication interface 150, and the processing circuit 160 are communicably connected to each other. Further, the device body 100 is connected to the network 50 .

The transmission/reception circuit 110 has a pulse generator, a transmission delay section, a pulser, etc., and supplies a drive signal to the ultrasonic probe 101 . A pulse generator repeatedly generates rate pulses at a predetermined rate frequency to form a transmitted ultrasound wave. In the transmission delay unit, the ultrasonic waves generated from the ultrasonic probe 101 are focused into a beam, and the pulse generator generates a delay time for each piezoelectric transducer necessary for determining the transmission directivity. given for each rate pulse. Also, the pulsar applies a driving signal (driving pulse) to the ultrasonic probe 101 at a timing based on the rate pulse. That is, the transmission delay unit arbitrarily adjusts the transmission direction of the ultrasonic waves transmitted from the piezoelectric transducer surface by changing the delay time given to each rate pulse.

The transmitting/receiving circuit 110 has a function capable of instantaneously changing the transmission frequency, the transmission drive voltage, etc. in order to execute a predetermined scan sequence based on instructions from the processing circuit 160, which will be described later. In particular, the change of the transmission drive voltage is realized by a linear amplifier type oscillator circuit capable of instantaneously switching the value or by a mechanism for electrically switching between a plurality of power supply units.

Further, the transmission/reception circuit 110 has a preamplifier, an A/D (Analog/Digital) converter, a reception delay unit, an adder, etc., and performs various processing on the reflected wave signal received by the ultrasonic probe 101 to Generate wave data. A preamplifier amplifies the reflected wave signal for each channel. The A/D converter A/D converts the amplified reflected wave signal. The reception delay section provides a delay time necessary for determining reception directivity. The adder adds the reflected wave signals processed by the reception delay unit to generate reflected wave data. The addition processing of the adder emphasizes the reflection component from the direction corresponding to the reception directivity of the reflected wave signal, and the reception directivity and the transmission directivity form a comprehensive beam for ultrasonic wave transmission/reception.

When the subject P is two-dimensionally scanned, the transmission/reception circuit 110 causes the ultrasonic probe 101 to transmit two-dimensional ultrasonic beams. Then, the transmission/reception circuit 110 generates two-dimensional reflected wave data from the two-dimensional reflected wave signal received by the ultrasonic probe 101 . Further, the transmission/reception circuit 110 according to the present embodiment causes the ultrasonic probe 101 to transmit a three-dimensional ultrasonic beam when three-dimensionally scanning the subject P. FIG. Then, the transmission/reception circuit 110 generates three-dimensional reflected wave data from the three-dimensional reflected wave signal received by the ultrasonic probe 101 .

Here, the form of the output signal from the transmitting/receiving circuit 110 can be various forms such as a signal containing phase information called an RF (Radio Frequency) signal, or amplitude information after envelope detection processing. It is selectable.

The B-mode processing circuit 120 receives the reflected wave data from the transmitting/receiving circuit 110, performs logarithmic amplification, envelope detection processing, etc., and generates data (B-mode data) in which the signal strength is represented by the brightness of luminance. .

The Doppler processing circuit 130 frequency-analyzes the velocity information from the reflected wave data received from the transmission/reception circuit 110, extracts blood flow, tissue, and contrast agent echo components due to the Doppler effect, and obtains moving body information such as velocity, dispersion, and power. Data extracted from multiple points (Doppler data) is generated. Specifically, the Doppler processing circuit 130 generates Doppler data at each of a plurality of sample points, such as average velocity, average dispersion value, average power value, etc., as motion information of the moving object. Here, the moving body is, for example, blood flow, tissue such as a heart wall, and a contrast agent. The Doppler processing circuit 130 uses information obtained by estimating the average velocity of blood flow, the average variance value of blood flow, the average power value of blood flow, etc. at each of a plurality of sample points as blood flow motion information (blood flow information). Generate.

The Doppler processing circuit 130 has an MTI (Moving Target Indicator) filter and a blood flow information generator. The Doppler processing circuit 130 performs, for example, a color Doppler method to calculate blood flow information. In the color Doppler method, ultrasonic waves are transmitted and received multiple times on the same scanning line, and by applying an MTI filter to the data train at the same position, the image is derived from stationary or slow-moving tissues. Signals (clutter signals) are suppressed to extract signals derived from blood flow. In the color Doppler method, blood flow information such as blood flow velocity, blood flow dispersion, and blood flow power is estimated from this blood flow signal.

The MTI filter uses a filter matrix to output a data string in which clutter components are suppressed and a blood flow signal derived from blood flow is extracted from a data string of continuous reflected wave data at the same position (same sample point). do. The blood flow information generator performs calculations such as autocorrelation using the data output from the MTI filter to estimate blood flow information, and outputs the estimated blood flow information as Doppler data. Examples of the MTI filter include a Butterworth IIR (Infinite Impulse Response) filter, a filter with fixed coefficients such as a polynomial regression filter, or a filter whose coefficients are fixed according to the input signal using an eigenvector or the like. An adaptive filter can be applied that changes .

The B-mode processing circuit 120 and the Doppler processing circuit 130 illustrated in FIG. 2 can process both two-dimensional reflected wave data and three-dimensional reflected wave data. That is, the B-mode processing circuit 120 generates two-dimensional B-mode data from the two-dimensional reflected wave data, and generates three-dimensional B-mode data from the three-dimensional reflected wave data. The Doppler processing circuit 130 also generates two-dimensional Doppler data from the two-dimensional reflected wave data, and generates three-dimensional Doppler data from the three-dimensional reflected wave data.

The storage circuit 140 is a memory that stores image data for display generated by the processing circuit 160 . The storage circuit 140 can also store data generated by the B-mode processing circuit 120 and the Doppler processing circuit 130 . The B-mode data and Doppler data stored in the storage circuit 140 can be called up by the operator after diagnosis, for example, and become ultrasonic image data for display via the processing circuit 160 .

In addition, the storage circuit 140 stores control programs for transmitting and receiving ultrasonic waves, image processing, and display processing, diagnostic information (for example, patient ID, doctor's findings, etc.), various data such as diagnostic protocols and various body marks. Remember. Data stored in the storage circuit 140 can be transferred to an external device via an interface (not shown). Note that the external device is, for example, a PC (Personal Computer) used by a doctor who performs image diagnosis, a storage medium such as a CD or DVD, a printer, or the like.

The communication interface 150 is an interface for communicating with various external devices via the network 50 . Communication interface 150 allows processing circuitry 160 to communicate with external devices. For example, the processing circuit 160 can exchange various data with an external device other than the medical image generating apparatus 10 via the communication interface 150 .

The processing circuitry 160 controls the overall processing of the medical image generating apparatus 10 . For example, processing circuitry 160 is implemented by a processor. A processing circuit 160 according to the first embodiment executes a control function 161 , an image generation function 162 , a setting function 163 , a first learning function 164 , an evaluation processing function 165 and a first change function 166 .

Here, for example, the control function 161, the image generation function 162, the setting function 163, the first learning function 164, the evaluation processing function 165, and the first change function 166, which are the components of the processing circuit 160 shown in FIG. Each processing function is stored in the memory circuit 140 in the form of a computer-executable program. The processing circuit 160 is a processor that reads each program from the storage circuit 140 and executes it, thereby implementing functions corresponding to each program. In other words, the processing circuit 160 with each program read has each function shown in the processing circuit 160 of FIG.

All the processing functions of the control function 161, the image generation function 162, the setting function 163, the first learning function 164, the evaluation processing function 165, and the first change function 166 are in the form of one program that can be executed by a computer, It may be recorded in the memory circuit 140 . For example, such programs are also referred to as medical image generation programs. In this case, the processing circuit 160 reads the medical image generation program from the storage circuit 140 and executes the read medical image generation program to perform control function 161, image generation function 162, setting function 163, and control function 161 corresponding to the medical image generation program. A first learning function 164, an evaluation processing function 165, and a first changing function 166 are realized.

The image generation function 162 is an example of an acquisition unit. The evaluation processing function 165 and the estimation processing function 169 are examples of processing units. The first change function 166 and the second change function 170 are examples of decision units. The setting function 163 is an example of a setting unit. First learning function 164 and second learning function 168 are examples of a learning unit. The first input function 167 and the second input function 171 are examples of input units.

The control function 161 executes overall control of the medical image generating apparatus 10 . For example, the control function 161 controls the communication interface 150 to acquire image data from the medical image generating apparatus 10 or PACS 30 . Then, the control function 161 causes the storage circuit 140 to store the acquired image data. Also, the control function 161 controls the input interface 102 and receives various operations.

The image generation function 162 acquires image data, which is an image including the subject P. FIG. More specifically, the image generating function 162 acquires ultrasound image data by generating or the like from the data generated by the B-mode processing circuit 120 and the Doppler processing circuit 130 based on the set parameters. That is, the image generation function 162 generates B-mode image data representing the intensity of the reflected wave in luminance from the B-mode data generated by the B-mode processing circuit 120 . Also, the image generation function 162 generates color Doppler image data as an average velocity image, a variance image, a power image, or a combination of these images representing moving body information from the Doppler data generated by the Doppler processing circuit 130 . That is, the image generation function 162 generates ultrasound image data that expresses the blood flow in the body of the subject P in colors based on the set parameters.

Here, the image generating function 162 generally converts (scan-converts) a scanning line signal train of ultrasonic scanning into a scanning line signal train of a video format typified by television and the like, and converts the ultrasonic waves for display. Generate image data. Specifically, the image generating function 162 generates ultrasonic image data for display by performing coordinate transformation according to the scanning mode of ultrasonic waves by the ultrasonic probe 101 . Also, the image generation function 162 synthesizes character information of various parameters, scales, body marks, various markers, etc. with the ultrasound image data.

The setting function 163 sets one or more regions of interest R1 for image data. More specifically, the setting function 163 sets the region of interest R1, which is the region to be evaluated by the trained model in the ultrasound image data. Here, FIG. 3 is an explanatory diagram for explaining the region of interest R1 set in the ultrasound image data. Since the inside of a body surface such as an organ of a subject P is often diagnosed in ultrasonic diagnosis, there are cases where it is desired to exclude the body surface from evaluation targets. In addition, since the contents to be drawn differ depending on the part to be diagnosed, that is, the depth indicating the depth from the body surface of the subject P, the parameters to be adjusted differ depending on the depth of the subject P. ing. For example, there are many cases where it is desired to draw fine blood vessels in the periphery of an area that is shallow from the body surface. In this case, the parameters are adjusted so as to draw a slow blood flow. On the other hand, in many cases, it is desired to draw thick blood vessels in areas deep from the body surface. In this case, if the parameters are adjusted so as to render slow blood flow, various movements are erroneously detected as blood flow, resulting in a noisy image.

For this reason, the setting function 163 sets the region of interest R1 for each depth. FIG. 3 shows a state in which the region of interest R1 is divided into three regions according to the depth of the subject P, for example. For example, the setting function 163 sets a first region of interest R11 from the lower end of the body surface to an arbitrary depth, a second region of interest R12 from the first region of interest R11 to an arbitrary depth, and an arbitrary depth from the second region of interest R12. A third region of interest R13 is set up to the depth. It should be noted that dividing the region of interest R1 into three is an example, and it may be divided into two or less, or may be divided into four or more. Also, the region of interest R1 is not limited to a rectangular shape, and may be of any shape.

Furthermore, the organs to be diagnosed are different in size. Also, the size of the displayed organ differs according to the angle of the ultrasonic probe 101 . Therefore, it is preferable that the region of interest R<b>1 differs according to the site to be diagnosed and the angle of the ultrasonic probe 101 . Here, the depth at which a diagnostic target site such as an organ is displayed and the size of the diagnostic target to be displayed may vary depending on the diagnostic target site, the position of the subject P to which the ultrasonic probe 101 is applied, and the ultrasonic wave. It can be estimated from the angle of the acoustic probe 101 .

Therefore, the setting function 163 is used to set the site to be diagnosed set before the ultrasonic probe 101 is applied, the position of the subject P to which the ultrasonic probe 101 is applied, the angle of the ultrasonic probe 101, and the like. One or more regions of interest R1 are set for each combination of conditions. That is, the setting function 163 sets the combination of these conditions to the depth determined based on the combination of the part to be diagnosed, the position of the subject P to which the ultrasonic probe 101 is applied, and the angle of the ultrasonic probe 101. A region of interest R1 having a size determined based on is set. At this time, the region of interest R1 is not limited to a rectangular shape, and may have another shape. Also, the angle of the ultrasonic probe 101 may be detected by adding a tilt sensor to the ultrasonic probe 101, or may be detected by another method. Then, the setting function 163 sets the region of interest R1 from the lower end of the body surface.

The first learning function 164 builds a learned model by learning the correlation between image data and evaluation values. Here, FIG. 4 is a diagram showing learning processing by the first learning function 164. As shown in FIG. The first learning function 164 inputs learning data obtained by combining ultrasonic image data generated with arbitrary parameters and evaluation values of the ultrasonic image data. Then, the first learning function 164 constructs a learned model for each region of interest R1 by learning the correlation between the ultrasound image data and the evaluation value of the ultrasound image data by machine learning. That is, the first learning function 164 constructs a learned model for each region of interest R1 set for each depth of the subject P. FIG. For example, the first learning function 164 builds trained models for each of a first region of interest R11, a second region of interest R12, and a third region of interest R13. The learned model constructed in this way outputs an evaluation value of the ultrasonic image data when the ultrasonic image data is input.

Here, the learning data in which the ultrasound image data and the evaluation value of the ultrasound image data are combined may be received from another device such as the medical information processing device 40, or may be received by the medical image generation device 10. may be generated. When generated by the medical image generating apparatus 10, the first learning function 164 generates, for example, by receiving an operation of inputting an evaluation value of ultrasonic image data for each ultrasonic image data.

Moreover, the first learning function 164 inputs an evaluation value for each evaluation item when an evaluation item is defined for each region of interest R1. For example, evaluation items include whether or not fine blood vessels in peripheral parts can be visualized, whether blood flow in deep parts can be captured, and how blood vessels can be connected. Moreover, the evaluation items may be different for each region of interest R1, or may be the same.

Furthermore, the first learning function 164 determines the region of interest R1 for each combination of conditions such as the site to be diagnosed, the position of the subject P to which the ultrasonic probe 101 is applied, and the angle of the ultrasonic probe 101. If set, a learned model is constructed for each of the one or more regions of interest R1 set for each combination of these conditions.

The evaluation processing function 165 obtains output information according to the image data acquired by the image generation function 162 using the learned model. That is, the evaluation processing function 165 obtains the evaluation value corresponding to the learning data acquired by the image generation function 162 as output information using a trained model that has learned the correlation between the image data and the evaluation value. Here, FIG. 5 is a diagram showing operation processing by the evaluation processing function 165. As shown in FIG. When the ultrasonic image data generated by the image generation function 162 with arbitrary parameters is input to the trained model corresponding to the region of interest R1 to be evaluated, the trained model includes the evaluation value of the input ultrasonic image data. Output output information. That is, the trained model outputs the evaluation value of the region of interest R1 to be evaluated in the input ultrasound image data. For example, a trained model targeting the first region of interest R11 outputs an evaluation value of the first region of interest R11. A trained model targeting the second region of interest R12 outputs an evaluation value of the second region of interest R12. A trained model targeting the third region of interest R13 outputs an evaluation value of the third region of interest R13. Furthermore, the learned model outputs an evaluation value for each evaluation item when an evaluation item is defined for each region of interest R1.

Then, the evaluation processing function 165 feeds back the evaluation value to the parameter. That is, the evaluation processing function 165 requests the first change function 166 to change the parameter when the evaluation value is less than the threshold. Here, when the region of interest R1 is set based on the site to be diagnosed, the position of the subject P to which the ultrasonic probe 101 is applied, and the angle of the ultrasonic probe 101, the evaluation processing function 165 identifies trained models that meet these conditions. Then, the evaluation processing function 165 inputs the ultrasound image data generated by the image generation function 162 to the identified trained model, and outputs output information including the evaluation value of the input ultrasound image data. Let

The first change function 166 determines parameters related to image data acquisition based on the output information obtained by the evaluation processing function 165 . In the first embodiment, the first change function 166 changes the parameter when the evaluation value included in the output information output by the trained model is less than the threshold. The first change function 166 changes the parameter in order to increase the evaluation value of the evaluation item with the lowest evaluation value when evaluation items are defined for each region of interest R1. For example, the first change function 166 shifts the MTI filter or Wall filter to the low-flow side or adjusts the flow-velocity range to be displayed when low-flow-velocity blood vessels such as peripheral fine blood vessels cannot be visualized. , so that even lower flow velocities can be captured. Conversely, if a lot of noise occurs in the ROI, the first change function 166 shifts the filter to the high flow velocity side to adjust the noise. In addition, the first change function 166 adjusts to increase the maximum detectable speed by lowering the transmission frequency when a high-flow blood vessel such as a deep blood vessel cannot be visualized.

Next, in the first embodiment, image generation processing when using a learned master will be described. FIG. 6 is a flow chart showing a processing procedure of image generation processing according to the first embodiment.

The control function 161 receives an operation for setting the part and depth to be diagnosed (step S11).

The image generation function 162 generates ultrasound image data based on the parameters (step S12).

The evaluation processing function 165 inputs the generated ultrasound image data to the trained model of the set depth, and outputs output information (step S13). That is, the evaluation processing function 165 obtains an evaluation value of ultrasonic image data for each evaluation item.

The first change function 166 determines whether the evaluation value of each evaluation item is less than the threshold (step S14).

If there is an evaluation value less than the threshold (step S14; Yes), the first change function 166 changes the parameter for increasing the evaluation value of the evaluation item with the lowest evaluation value (step S15). Then, the medical image generating apparatus 10 proceeds to step S12 and generates ultrasonic image data again.

On the other hand, if there is no evaluation value less than the threshold (step S14; No), the image generation function 162 generates ultrasound image data based on the parameters (step S16).

The control function 161 determines whether or not an operation to terminate acquisition of ultrasound image data has been received (step S17). If an operation to terminate acquisition of ultrasound image data has not been received (step S17; No), the medical image generating apparatus 10 proceeds to step S16 and generates ultrasound image data again.

If an operation to terminate the acquisition of ultrasound image data has been received (step S17; Yes), the medical image generating apparatus 10 terminates the image generation processing.

As described above, the medical image generating apparatus 10 according to the first embodiment acquires the evaluation value of the input ultrasound image data by inputting the ultrasound image data into the trained model. Then, when the evaluation value is less than the threshold, the medical image generating apparatus 10 changes the parameters and generates ultrasound image data again. By repeatedly executing this process, the medical image generating apparatus 10 can set appropriate parameters and generate ultrasound image data with an evaluation value equal to or greater than the threshold. That is, the medical image generating apparatus 10 can improve the image quality of the generated ultrasound image data.

(First Modification of First Embodiment)
A first modification of the first embodiment will be described. Here, FIG. 7 is a block diagram showing an example of the configuration of the medical image generating apparatus 10a according to the first modification of the first embodiment. The processing circuit 160a of the medical image generating apparatus 10a according to the first modification of the first embodiment has a first input function 167 instead of the first learning function 164. FIG.

The first input function 167 accepts an input such as by receiving a trained model from another device such as the medical information processing device 40 . More specifically, the first input function 167 receives an input of a trained model constructed by learning the correlation between ultrasound image data and evaluation values. Note that the first input function 167 is not limited to receiving, and may receive input from a storage medium storing a learned model via a connection interface.

As described above, the medical image generating apparatus 10a according to the first modification of the first embodiment accepts inputs of trained models. Therefore, the medical image generating apparatus 10a can acquire a trained model without having the first learning function 164. FIG. Therefore, the medical image generating apparatus 10a can generate image data with appropriate parameters without the first learning function 164. FIG. That is, the medical image generating apparatus 10a can improve the image quality of the generated ultrasound image data even without the first learning function 164 .

(Second embodiment)
A second embodiment will be described. Here, FIG. 8 is a block diagram showing an example of the configuration of the medical image generating apparatus 10b according to the second embodiment. The processing circuit 160 b of the medical information processing apparatus 40 according to the second embodiment has a second learning function 168 instead of the first learning function 164 . The medical information processing apparatus 40 also has an estimation processing function 169 instead of the evaluation processing function 165 . The medical information processing apparatus 40 also has a second change function 170 instead of the first change function 166 .

A second learning function 168 builds a trained model by learning correlations between image data and parameters. Here, FIG. 9 is a diagram showing learning processing by the second learning function 168. As shown in FIG. The second learning function 168 constructs a learned model by machine learning using ultrasonic image data and appropriate parameters that, when set, make the evaluation value of the ultrasonic image data equal to or higher than the threshold. More specifically, the second learning function 168 inputs learning data obtained by combining ultrasound image data generated by the image generating function 162 with default parameters and appropriate parameters. Then, the second learning function 168 builds a learned model for each region of interest R1 by learning the correlation between the ultrasound image data and appropriate parameters through machine learning. That is, the second learning function 168 constructs a learned model for each region of interest R1 set for each depth of the subject P. FIG. For example, the second learning function 168 builds trained models for each of a first region of interest R11, a second region of interest R12, and a third region of interest R13. Furthermore, the second learning function 168 is used when the region of interest R1 is set based on the site to be diagnosed, the position of the subject P to which the ultrasonic probe 101 is applied, and the angle of the ultrasonic probe 101. constructs a trained model for each of one or more regions of interest R1 set for each combination of these conditions. When ultrasonic image data is input, the trained model generated in this way can be reset to generate ultrasonic image data, thereby generating ultrasonic image data with an evaluation value equal to or greater than the threshold. Output the appropriate parameters that can be

Here, the learning data in which the ultrasound image data and appropriate parameters are combined may be received from another device such as the medical information processing device 40, or may be generated by the medical image generation device 10b. good. When generated by the medical image generating apparatus 10b, the second learning function 168 generates by accepting an operation of inputting appropriate parameters for each ultrasonic image data.

The estimation processing function 169 outputs the parameter corresponding to the learning data acquired by the image generation function 162 using a trained model that has learned the correlation between the image data and the parameter that makes the evaluation value of the image data equal to or higher than the threshold. get information. That is, the estimation processing function 169 uses a trained model that has learned the correlation between the image data and the parameter that makes the evaluation value of the image data equal to or higher than the threshold, and uses the parameter corresponding to the image data acquired by the image generation function 162. is obtained as output information. Here, FIG. 10 is a diagram showing operation processing by the estimation processing function 169. As shown in FIG. The estimation processing function 169 inputs the ultrasound image data generated by the image generation function 162 to the trained model, thereby generating ultrasound image data whose evaluation value is equal to or greater than a threshold. output information that includes . That is, the trained model outputs appropriate parameters for the input ultrasound image data. At this time, the estimation processing function 169 inputs the ultrasound image data generated by the image generation function 162 with the default parameters to the trained model corresponding to the region of interest R1 to be evaluated. For example, when the first region of interest R11 is to be evaluated, the estimation processing function 169 inputs ultrasound image data to the trained model of the first region of interest R11. When the second region of interest R12 is to be evaluated, the estimation processing function 169 inputs ultrasound image data to the trained model of the second region of interest R12. When the third region of interest R13 is to be evaluated, the estimation processing function 169 inputs ultrasound image data to the trained model of the third region of interest R13. Then, each trained model outputs its parameters.

Here, if the region of interest R1 is set based on the site to be diagnosed, the position of the subject P to which the ultrasonic probe 101 is applied, or the angle of the ultrasonic probe 101, the estimation processing function 169 identifies trained models that meet these conditions. Then, the estimation processing function 169 inputs the ultrasonic image data generated by the image generating function 162 to the identified trained model, thereby making the evaluation value of the input ultrasonic image data equal to or higher than the threshold. Causes output information containing parameters to be output.

The second change function 170 determines parameters for the image generation function 162 to generate ultrasound image data based on the output information that the estimation processing function 169 causes the trained model to output. In the second embodiment, the second change function 170 changes the parameter to be included in the output information output by the learned model. Appropriate ultrasonic image data is generated by generating ultrasonic image data using the parameters output by the image generating function 162 . Here, appropriate ultrasound image data is ultrasound image data whose evaluation value is equal to or greater than a threshold.

Next, in the second embodiment, image generation processing when using a learned master will be described. FIG. 11 is a flow chart showing a processing procedure of image generation processing according to the second embodiment.

The control function 161 receives an operation for setting the part and depth to be diagnosed (step S21).

The image generation function 162 generates ultrasound image data based on the parameters (step S22).

The estimation processing function 169 inputs the generated ultrasound image data to the trained model of the set depth and outputs output information (step S23). That is, the estimation processing function 169 outputs appropriate parameters to be set in generating the input ultrasound image data.

The second change function 170 changes the setting to the output parameter (step S24).

The image generation function 162 generates ultrasound image data based on the parameters (step S25). That is, the image generation function 162 generates appropriate ultrasound image data based on appropriate parameters.

The control function 161 determines whether or not an operation to terminate acquisition of ultrasound image data has been received (step S26). If an operation to terminate acquisition of ultrasound image data has not been received (step S26; No), the medical image generating apparatus 10b proceeds to step S25 and generates ultrasound image data again.

When receiving an operation to terminate the acquisition of ultrasound image data (step S26; Yes), the medical image generating apparatus 10b terminates the image generation processing.

As described above, the medical image generation apparatus 10b according to the second embodiment acquires appropriate parameters to be set in generating the input ultrasound image data by inputting the ultrasound image data into the trained model. do. Therefore, the medical image generating apparatus 10b can generate image data with appropriate parameters. That is, the medical image generating apparatus 10b can improve the image quality of the generated ultrasound image data.

(First Modification of Second Embodiment)
A first modification of the second embodiment will be described. Here, FIG. 12 is a block diagram showing an example of the configuration of a medical image generating apparatus 10c according to the first modified example of the second embodiment. The processing circuit 160c of the medical image generating apparatus 10c according to the first modification of the second embodiment has a second input function 171 instead of the second learning function 168. FIG.

The second input function 171 accepts input such as by receiving a trained model from another device such as the medical information processing device 40 . More specifically, the second input function 171 receives input of ultrasound image data and parameters that make the evaluation value of the ultrasound image data equal to or higher than a threshold, and learns the correlation between the ultrasound image data and the parameters. Accepts the input of a trained model built by In addition, the second input function 171 may receive input from a storage medium storing a learned model via a connection interface instead of receiving.

As described above, the medical image generating apparatus 10c according to the first modification of the second embodiment accepts inputs of trained models. Therefore, the medical image generating apparatus 10c can acquire a trained model without the second learning function 168. FIG. Therefore, the medical image generating apparatus 10c can generate image data with appropriate parameters without the second learning function 168. FIG. That is, the medical image generating apparatus 10c can improve the image quality of the generated ultrasound image data even without the second learning function 168 .

Further, in the above-described embodiment, an example in which each processing function is realized by a single processing circuit (processing circuits 160, 160a, 160b, 160c) has been described, but the embodiment is not limited to this. For example, the processing circuits 160, 160a, 160b, and 160c may be configured by combining a plurality of independent processors, and each processor may implement each processing function by executing each program. Moreover, each processing function of the processing circuits 160, 160a, 160b, and 160c may be appropriately distributed or integrated in a single or a plurality of processing circuits and implemented.

The word "processor" used in the description of each embodiment described above is, for example, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or an application specific integrated circuit (ASIC), a programmable Means circuits such as logic devices (for example, Simple Programmable Logic Device (SPLD), Complex Programmable Logic Device (CPLD), and Field Programmable Gate Array (FPGA)) do. Here, instead of storing the program in the memory, the program may be configured to be directly embedded in the circuit of the processor. In this case, the processor implements its functions by reading and executing the program embedded in the circuit. Further, each processor of the present embodiment is not limited to being configured as a single circuit for each processor, and may be configured as one processor by combining a plurality of independent circuits to realize its function. good.

Here, the program executed by the processor is preinstalled in a ROM (Read Only Memory), storage unit, or the like and provided. This program is a file in a format that can be installed in these devices or in a format that can be executed, such as CD (Compact Disk)-ROM, FD (Flexible Disk), CD-R (Recordable), DVD (Digital Versatile Disk), etc. may be provided on a computer readable storage medium. Also, this program may be provided or distributed by being stored on a computer connected to a network such as the Internet and downloaded via the network. For example, this program is composed of modules containing each functional unit. As actual hardware, the CPU reads out a program from a storage medium such as a ROM and executes it, so that each module is loaded onto the main storage device and generated on the main storage device.

Also, each component of each device illustrated is functionally conceptual, and does not necessarily need to be physically configured as illustrated. In other words, the specific form of distribution and integration of each device is not limited to the illustrated one, and all or part of them can be functionally or physically distributed and integrated in arbitrary units according to various loads and usage conditions. Can be integrated and configured. Furthermore, each processing function performed by each device may be implemented in whole or in part by a CPU and a program analyzed and executed by the CPU, or implemented as hardware based on wired logic.

Further, the medical image processing method described in the above embodiments can be realized by executing a prepared program on a computer such as a personal computer or a workstation. This program can be distributed via a network such as the Internet. The program can also be recorded on a computer-readable recording medium such as a hard disk, flexible disk (FD), CD-ROM, MO, DVD, etc., and executed by being read from the recording medium by a computer.

According to at least one embodiment described above, image data can be generated with appropriate parameters.

While several embodiments of the invention have been described, these embodiments have been presented by way of example and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, replacements, and modifications can be made without departing from the scope of the invention. These embodiments and their modifications are included in the scope and spirit of the invention, as well as the scope of the invention described in the claims and equivalents thereof.

1 medical information processing system 10, 10a, 10b, 10c medical image generation device 20 electronic chart system 30 PACS
40 medical information processing apparatus 50 network 160, 160a, 160b, 160c processing circuit 161 control function 162 image generation function 163 setting function 164 first learning function 165 evaluation processing function 166 first changing function 167 first input function 168 second learning function 169 estimation processing function 170 second change function 171 second input function

Claims (5)

an acquisition unit that acquires subject image data, which is an image including the subject;
a setting unit that sets at least two regions of interest having different depths of the subject for the subject image data;
The subject image data is input to a trained model that has learned the correlation between the image data and the evaluation value for each depth of the image data, and the evaluation value of the subject image data is indicated for each region of interest . a processing unit that obtains output information;
a determination unit that determines parameters related to acquisition of the subject image data so that the evaluation value indicated for each region of interest in the output information obtained by the processing unit satisfies a reference value ;
A medical image generation device comprising: The setting unit sets two or more regions of interest for each combination of a site to be diagnosed and an angle of the probe .
The medical image generating apparatus according to claim 1 . further comprising a first input unit that receives an input of the trained model constructed by learning the correlation between the subject image data and the evaluation value;
The medical image generating apparatus according to claim 1 or 2 . The determination unit determines parameters related to the subject image data in which blood vessels are drawn according to the depth, based on the output information obtained by the processing unit.
The medical image generation apparatus according to any one of claims 1 to 3 . acquiring subject image data, which is an image including the subject;
setting at least two regions of interest having different depths of the subject for the subject image data;
The subject image data is input to a trained model that has learned the correlation between the image data and the evaluation value for each depth of the image data, and the evaluation value of the subject image data is indicated for each region of interest . get the output information,
Determining a parameter related to acquisition of the subject image data so that the evaluation value indicated for each region of interest of the obtained output information satisfies a reference value ;
A medical image generation program that causes a computer to execute processing.

Images