Medical image diagnostic device, method for determining medical imaging conditions, and program

Abstract

To acquire a medical image that enables highly accurate diagnosis to be executed while causing imaging and analyzing to influence each other.SOLUTION: A medical image diagnostic device includes an acquisition unit and a determination unit. The acquisition unit acquires a medical image of a subject. When the medical image is input, the determination unit determines an imaging condition for the medical image of the subject on the basis of a model learned to output an examination result of a person included in the input medical image, and a map in which a degree of contribution to the examination result is associated with each position on the input medical image, and the medical image of the subject acquired by the acquisition unit.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The embodiments disclosed in this specification and the drawings relate to a medical image diagnostic apparatus, a method for determining medical imaging conditions, and a program. [Background technology]

[0002] In today's clinical settings, imaging staff optimize imaging protocols to obtain the highest quality medical images. Analysts attempt to obtain clinical information from the medical images they are given. In other words, medical image capture and analysis tend to be performed independently of each other. For this reason, even if artificial intelligence (AI) for analyzing medical images makes significant advances in the future, it is expected that this will have little impact on the imaging side.

[0003] Meanwhile, research and development and clinical application of image analysis AI, which inputs medical images and outputs diagnostic information, is progressing. In this context, a technology is known that estimates which parts of the input image contribute most to the final output. [Prior art documents] [Non-patent literature]

[0004] [Non-Patent Document 1] Fei Wang, Mengqing Jiang, Chen Qian, Shuo Yang, Cheng Li, Honggang Zhang, Xiaogang Wang, Xiaoou Tang “Residual Attention Network for Image Classification”, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 21-26 July 2017, DOI 10.1109 / CVPR.2017.683 [Non-patent document 2] Grzegorz Jacenkow, Alison Q. O'Neil, Brian Mohr, Sotirios A. Tsaftaris “Steering Spatial Attention with Non-Imaging Information in CNNs”, Medical Image Computing and Computer Assisted Intervention - MICCAI 2020, 23rd International Conference, Lima, Peru, October 4-8, 2020, Proceedings, Part IV (pp.385-395), DOI 10.1007 / 978-3-030-59719-1_38 Summary of the Invention [Problem to be solved by the invention]

[0005] The problem to be solved by the embodiments disclosed in this specification and the drawings is to obtain medical images that enable highly accurate diagnosis while allowing imaging and analysis to interact with each other. However, the problem to be solved by the embodiments disclosed in this specification and the drawings is not limited to the above problem. Problems corresponding to the effects of each configuration shown in the embodiments described below can also be positioned as other problems. [Means for solving the problem]

[0006] According to an embodiment, a medical image diagnostic apparatus includes an acquisition unit and a determination unit. The acquisition unit acquires a medical image of a subject. When the medical image is input, the determination unit determines imaging conditions for the medical image of the subject based on a model trained to output a medical result of a person included in the input medical image and a map in which each position on the input medical image is associated with a contribution to the medical result, and the medical image of the subject acquired by the acquisition unit. [Brief explanation of the drawings]

[0007] [Figure 1]1 is a diagram showing an example of the configuration of a medical image diagnostic system 1 according to an embodiment. [Figure 2] 1 is a diagram showing an example of the arrangement of an X-ray CT apparatus 100 according to an embodiment. [Figure 3] FIG. 2 is a diagram illustrating an example of the configuration of a learning device 200 according to an embodiment. [Figure 4] 3 is a flowchart showing the flow of a series of processes during training by the medical image diagnostic system 1 according to the embodiment. [Figure 5] FIG. 10 is a diagram for explaining a method for obtaining a contribution map during training. [Figure 6] 3 is a flowchart showing the flow of a series of processes at runtime by the medical image diagnostic system 1 according to the embodiment. [Figure 7] FIG. 10 is a diagram schematically illustrating a method for determining imaging conditions using a contribution map. [Figure 8] FIG. 10 is a diagram schematically illustrating a method for determining imaging conditions using a contribution map. [Figure 9] FIG. 10 is a diagram schematically illustrating a method for determining imaging conditions using a contribution map. [Figure 10] FIG. 10 is a diagram for explaining a method of aligning contribution maps. [Figure 11] FIG. 10 is a diagram for explaining a method of aligning contribution maps. [Figure 12] FIG. 10 is a diagram for explaining a method of aligning contribution maps. [Figure 13] FIG. 10 is a diagram for explaining a method of aligning contribution maps. [Figure 14] FIG. 10 is a diagram for explaining a method for determining imaging conditions. [Figure 15] FIG. 10 is a diagram for explaining a method for determining imaging conditions. [Figure 16] FIG. 10 is a diagram for explaining a method for determining imaging conditions. [Figure 17] FIG. 10 is a diagram for explaining a method for determining imaging conditions. [Figure 18] FIG. 10 is a diagram for explaining a method for determining imaging conditions. [Figure 19]FIG. 10 is a diagram for explaining a method for determining imaging conditions. [Figure 20] FIG. 10 is a diagram for explaining a method for determining imaging conditions. [Figure 21] FIG. 10 is a diagram for explaining a method for determining imaging conditions. [Figure 22] FIG. 10 is a diagram for explaining a method for determining imaging conditions. [Figure 23] FIG. 10 is a diagram for explaining a method for determining imaging conditions. [Figure 24] FIG. 10 is a diagram for explaining a method for determining imaging conditions. [Figure 25] FIG. 10 is a diagram for explaining a method for determining imaging conditions. [Figure 26] FIG. 10 is a diagram for explaining a method for determining imaging conditions. [Figure 27] FIG. 10 is a diagram for explaining a method for determining imaging conditions. [Figure 28] FIG. 10 is a diagram for explaining a method for determining imaging conditions. DETAILED DESCRIPTION OF THE INVENTION

[0008] Hereinafter, a medical image diagnostic apparatus, a method for determining medical imaging conditions, and a program according to an embodiment will be described with reference to the drawings.

[0009] [Configuration of medical image diagnostic system] 1 is a diagram illustrating an example of the configuration of a medical image diagnostic system 1 according to an embodiment. The medical image diagnostic system 1 includes, for example, one or more medical image capturing devices 100 and a learning device 200. The medical image capturing devices 100 and the learning device 200 are communicably connected via a communication network NW.

[0010] The term "communication network NW" refers to any information and communication network that utilizes telecommunications technology. It includes wireless / wired LANs such as hospital backbone LANs (Local Area Networks) and the Internet, as well as telephone communication networks, optical fiber communication networks, cable communication networks, and satellite communication networks.

[0011] The medical imaging apparatus 100 is an apparatus that generates a medical image by scanning a subject P. The subject P is, for example, a human patient, but is not limited to this, and may be an animal or an inorganic object. The medical imaging apparatus 100 may be, for example, an X-ray CT (Computed Tomography) apparatus, or may also be an MRI (Magnetic Resonance Imaging) apparatus, an ultrasound imaging apparatus, a nuclear medicine diagnostic apparatus, or the like. The medical imaging apparatus 100 is a so-called modality. In the following, as an example, the medical imaging apparatus 100 will be described as an X-ray CT apparatus.

[0012] The learning device 200 collects medical images from an X-ray CT apparatus (medical imaging apparatus) 100, learns a machine learning model using the collected medical images, and analyzes the medical images using the learned model. Details of the model will be described later. The learning device 200 may be a workstation or a cloud server connected to the X-ray CT apparatus 100, which is a modality.

[0013] [X-ray CT system configuration] FIG. 2 is a diagram illustrating an example of the configuration of an X-ray CT apparatus 100 according to an embodiment. The X-ray CT apparatus 100 includes, for example, a gantry 110, a bed 130, and a console 140. For convenience of explanation, FIG. 2 illustrates the gantry 110 viewed from both the Z-axis direction and the X-axis direction, but in reality, there is only one gantry 110. In this embodiment, the rotation axis of the rotating frame 117 in a non-tilted state or the longitudinal direction of the tabletop 133 of the bed 130 is defined as the Z-axis direction, the axis perpendicular to the Z-axis direction and horizontal to the floor surface is defined as the X-axis direction, and the direction perpendicular to the Z-axis direction and perpendicular to the floor surface is defined as the Y-axis direction.

[0014] The gantry device 110 includes, for example, an X-ray tube 111, a wedge 112, a collimator 113, an X-ray high voltage device 114, an X-ray detector 115, a data acquisition system (hereinafter referred to as DAS: Data Acquisition System) 116, a rotating frame 117, and a control device 118.

[0015] The X-ray tube 111 generates X-rays by irradiating thermoelectrons from a cathode (filament) to an anode (target) when a high voltage is applied from the X-ray high voltage device 114. The X-ray tube 111 includes a vacuum tube. For example, the X-ray tube 111 is a rotating anode type X-ray tube that generates X-rays by irradiating a rotating anode with thermoelectrons.

[0016] The wedge 112 is a filter for adjusting the amount of X-rays irradiated from the X-ray tube 111 to the subject P. The wedge 112 attenuates the X-rays passing through it so that the distribution of the amount of X-rays irradiated from the X-ray tube 111 to the subject P becomes a predetermined distribution. The wedge 112 is also called a wedge filter or a bow-tie filter. The wedge 112 is made by processing aluminum so as to have a predetermined target angle and a predetermined thickness, for example.

[0017] The collimator 113 is a mechanism for narrowing down the irradiation range of the X-rays that have passed through the wedge 112. The collimator 113 narrows down the irradiation range of the X-rays, for example, by forming a slit by combining multiple lead plates. The collimator 113 is also called an X-ray aperture.

[0018] The X-ray high voltage device 114 includes, for example, a high voltage generator and an X-ray control device. The high voltage generator has an electric circuit including a transformer, a rectifier, etc., and generates a high voltage to be applied to the X-ray tube 111. The X-ray control device controls the output voltage of the high voltage generator according to the X-ray dose to be generated in the X-ray tube 111. The high voltage generator may be one that performs voltage boosting using the above-mentioned transformer, or one that performs voltage boosting using an inverter. The X-ray high voltage device 114 may be provided on the rotating frame 117, or may be provided on the side of the fixed frame (not shown) of the gantry device 110.

[0019] The X-ray detector 115 detects the intensity of X-rays generated by the X-ray tube 111 and incident upon the subject P. The X-ray detector 115 outputs an electrical signal (which may be an optical signal, etc.) corresponding to the intensity of the detected X-rays to the DAS 116. The X-ray detector 115 has, for example, multiple X-ray detection element rows. Each of the multiple X-ray detection element rows has multiple X-ray detection elements arranged in the channel direction along an arc centered on the focal point of the X-ray tube 111. The multiple X-ray detection element rows are arranged in the slice direction (column direction, row direction).

[0020] The X-ray detector 115 is an indirect detector having, for example, a grid, a scintillator array, and a photosensor array. The scintillator array has multiple scintillators. Each scintillator has scintillator crystals. The scintillator crystals emit light with an amount of light corresponding to the intensity of the incident X-rays. The grid is arranged on the surface of the scintillator array on which the X-rays are incident and has an X-ray shielding plate that has the function of absorbing scattered X-rays. The grid is sometimes called a collimator (one-dimensional collimator or two-dimensional collimator). The photosensor array has, for example, a photosensor such as a photomultiplier tube (PMT). The photosensor array outputs an electrical signal corresponding to the amount of light emitted by the scintillator. The X-ray detector 115 may also be a direct conversion detector having a semiconductor element that converts incident X-rays into an electrical signal.

[0021] The DAS 116 includes, for example, an amplifier, an integrator, and an A / D converter. The amplifier amplifies the electrical signal output by each X-ray detection element of the X-ray detector 115. The integrator integrates the amplified electrical signal over a view period (described below). The A / D converter converts the electrical signal indicating the integration result into a digital signal. The DAS 116 outputs detection data based on the digital signal to the console device 140. The detection data is a digital value of X-ray intensity identified by the channel number and column number of the X-ray detection element that generated the data, and a view number indicating the acquired view. The view number is a number that changes according to the rotation of the rotating frame 117, and is, for example, a number that is incremented according to the rotation of the rotating frame 117. Therefore, the view number is information that indicates the rotation angle of the X-ray tube 111. The view period is the period from the rotation angle corresponding to a certain view number to the rotation angle corresponding to the next view number. The DAS 116 may detect the view switching by a timing signal input from the control device 118, by an internal timer, or by a signal acquired from a sensor (not shown). When a full scan is performed and X-rays are continuously emitted by the X-ray tube 111, the DAS 116 collects a group of detection data for the entire circumference (360 degrees). When a half scan is performed and X-rays are continuously emitted by the X-ray tube 111, the DAS 116 collects detection data for half the circumference (180 degrees).

[0022] The rotating frame 117 is an annular rotating member that rotates the X-ray tube 111, wedge 112, collimator 113, and X-ray detector 115 while holding them facing each other. The rotating frame 117 is supported by a fixed frame so as to be rotatable around the subject P introduced inside. The rotating frame 117 also supports the DAS 116. Detection data output by the DAS 116 is transmitted by optical communication from a transmitter having a light-emitting diode (LED) provided on the rotating frame 117 to a receiver having a photodiode provided on a non-rotating portion of the gantry device 110 (e.g., the fixed frame), and then transferred to the console device 140 by the receiver. Note that the method of transmitting the detection data from the rotating frame 117 to the non-rotating portion is not limited to the above-mentioned method using optical communication, and any non-contact transmission method may be adopted. The rotating frame 117 is not limited to an annular member, and may be an arm-like member as long as it can support and rotate the X-ray tube 111 and the like.

[0023] The control device 118 includes, for example, a processing circuit having a processor such as a CPU (Central Processing Unit), and a drive mechanism including a motor, an actuator, etc. The control device 118 receives input signals from an input interface 143 attached to the console device 140 or the gantry device 110, and controls the operations of the gantry device 110 and the bed device 130.

[0024] The control device 118, for example, rotates the rotating frame 117, tilts the gantry 110, or moves the tabletop 133 of the bed device 130. When tilting the gantry 110, the control device 118 rotates the rotating frame 117 around an axis parallel to the Z-axis direction based on the inclination angle (tilt angle) input to the input interface 143. The control device 118 grasps the rotation angle of the rotating frame 117 from the output of a sensor (not shown), etc. The control device 118 also provides the rotation angle of the rotating frame 117 to the processing circuit 150 as needed. The control device 118 may be provided in the gantry 110 or in the console device 140.

[0025] The control device 118 moves the gantry device 110 along the moving rails, and performs a main scan or a scanogram, which is a positioning scan performed before the main scan.

[0026] The bed device 130 is a device on which the subject P to be scanned is placed and introduced into the rotating frame 117 of the gantry device 110. The bed device 130 has, for example, a base 131, a bed driving device 132, a top plate 133, and a support frame 134. The base 131 includes a housing that supports the support frame 134 so that the support frame 134 can move in the vertical direction (Y-axis direction). The bed driving device 132 includes a motor and an actuator. The bed driving device 132 moves the top plate 133, on which the subject P is placed, along the support frame 134 in the longitudinal direction of the top plate 133 (Z-axis direction). The top plate 133 is a plate-shaped member on which the subject P is placed.

[0027] The console device 140 includes, for example, a memory 141, a display 142, an input interface 143, a communication interface 144, and a processing circuit 150. In this embodiment, the console device 140 is described as being separate from the gantry device 110, but the gantry device 110 may include some or all of the components of the console device 140. The console device 140 is an example of a "medical image diagnostic device."

[0028] The memory 141 is realized by, for example, a semiconductor memory element such as a random access memory (RAM) or a flash memory, a hard disk, an optical disk, etc. The memory 141 may also include a storage medium such as a read only memory (ROM) or a register. The memory 208 may also include a storage medium such as a read only memory (ROM) or a register.

[0029] The memory 141 stores, for example, detection data, projection data, reconstructed images, etc. These data may be stored in an external memory (for example, a NAS (Network Attached Storage)) with which the X-ray CT apparatus 100 can communicate, instead of (or in addition to) the memory 141.

[0030] Furthermore, the memory 141 stores model information. The model information is information (a program or a data structure) that defines the trained model MDL. The trained model MDL is typically a model that has been trained to output, when a medical image of a certain subject P is input, a medical examination result of the subject P and a contribution map.

[0031] The diagnosis result may indicate, for example, whether or not the subject P has a specific disease, or may indicate the probability of the specific disease (likelihood of the specific disease). For example, if the imaging site is the heart, the specific disease may be a heart valve disease.

[0032] Furthermore, the medical results are not limited to the above, and may represent, for example, measured values such as cancer growth rates or rehabilitation success rates, comparison values such as the degree of non-match between medical images and other tests, prediction results such as prognosis predictions or treatment side effects predictions, or derivative analysis results thereof. In other words, the medical results according to this embodiment may include any event that can be output by the trained model MDL.

[0033] A contribution map is two-dimensional or higher data in which the contribution to the medical outcome is associated with each position on the medical image input to the trained model MDL. For example, on a medical image, the greater the contribution, the greater the influence on the medical outcome. In other words, from the perspective of the trained model MDL, the greater the contribution, the more attention is paid to a position (area) that has a greater contribution when determining the medical outcome.

[0034] The trained model MDL may be implemented by, for example, a deep neural network (DNN) such as a convolutional neural network (CNN). Such a DNN includes an attention mechanism for obtaining the contribution map described above.

[0035] An attention mechanism is a neural network that changes the attention given to a pixel in a medical image depending on where that pixel is located in the entire image. For example, the attention mechanism multiplies each pixel on the feature map output from the convolutional layer of a CNN by a different weight, thereby determining the degree of attention (i.e., attention) for each pixel.

[0036] Furthermore, the trained model MDL is not limited to DNNs that include an attention mechanism, and may be implemented using other models or algorithms such as support vector machines, decision trees, naive Bayes classifiers, and random forests.

[0037] When the trained model MDL is implemented using a DNN, the model information includes, for example, connection information describing how the units (units or nodes) included in the input layer, one or more hidden layers (intermediate layers), and output layer of the DNN are connected to each other, and weight information describing the connection coefficients assigned to data exchanged between the connected units. The connection information includes, for example, the number of units included in each layer, information specifying the type of unit to which each unit is connected, activation functions that realize each unit, and gates provided between units in the hidden layers. The activation functions that realize the units may be, for example, a rectified linear unit (ReLU) function, an exponential linear unit (ELU) function, a clipping function, a sigmoid function, a step function, a hyperpolytangent function, or an identity function. The gates selectively pass or weight data transmitted between units depending on, for example, the value (e.g., 1 or 0) returned by the activation function. The connection coefficients include, for example, weights assigned to output data when data is output from a unit in a layer to a unit in a deeper layer in a hidden layer of a neural network. The connection coefficients may also include bias components specific to each layer.

[0038] The display 142 displays various types of information. For example, the display 142 displays a CT image generated by the processing circuitry 150, a GUI (Graphical User Interface) that accepts various operations by an operator (e.g., a doctor), etc. The display 142 is, for example, a liquid crystal display (LCD), a cathode ray tube (CRT) display, an organic electroluminescence (EL) display, etc. The display 142 may be provided on the gantry device 110. The display 142 may be a desktop type, or may be a display device (e.g., a tablet terminal) that can communicate wirelessly with the main body of the console device 140.

[0039] The input interface 143 accepts various input operations from an operator (e.g., a doctor) and outputs an electrical signal indicating the content of the accepted input operation to the processing circuit 150. For example, the input interface 143 accepts input operations such as an imaging protocol for imaging the subject P, reconstruction conditions for reconstructing CT images, and image processing conditions for generating post-processed images from CT images. For example, the input interface 143 is realized by a mouse, keyboard, touch panel, drag ball, switch, button, joystick, foot pedal, camera, infrared sensor, microphone, etc. The input interface 143 may be provided in the gantry device 110. The input interface 143 may also be realized by a display device (e.g., a tablet terminal) capable of wireless communication with the main body of the console device 140. Note that, in this specification, the input interface 143 is not limited to one having physical operation components such as a mouse and keyboard. For example, the input interface 143 also 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 a control circuit.

[0040] The communication interface 144 includes, for example, a network interface card (NIC), a wireless communication module, etc. The communication interface 144 communicates with the learning device 200 and other external devices via the communication network NW.

[0041] The processing circuitry 150 controls the overall operation of the X-ray CT apparatus 100. The processing circuitry 150 executes, for example, a system control function 151, an imaging condition determination function 152, a pre-processing function 153, a reconstruction processing function 154, an image processing function 155, an output control function 156, and the like. The processing circuitry 150 realizes these functions by, for example, a hardware processor executing a program stored in the memory 141. The pre-processing function 153 and the reconstruction processing function 154 are examples of an "acquisition unit." The imaging condition determination function 152 is an example of a "determination unit."

[0042] A hardware processor refers to a circuit such as a CPU, a graphics processing unit (GPU), an application specific integrated circuit (ASIC), a programmable logic device (e.g., a simple programmable logic device (SPLD) or a complex programmable logic device (CPLD), or a field programmable gate array (FPGA)).

[0043] Instead of storing a program in the memory 141, the program may be directly embedded in the circuit of the hardware processor. In this case, the hardware processor realizes the function by reading and executing the program embedded in the circuit. The program may be stored in the memory 141 in advance, or may be stored in a non-transitory storage medium such as a DVD or CD-ROM, and installed from the non-transitory storage medium into the memory 141 when the non-transitory storage medium is inserted into a drive device (not shown) of the console device 140. The hardware processor is not limited to being configured as a single circuit, and may be configured as a single hardware processor by combining multiple independent circuits to realize each function. Furthermore, multiple components may be integrated into a single hardware processor to realize each function.

[0044] Each component of the console device 140 or the processing circuitry 150 may be distributed and realized by multiple pieces of hardware. The processing circuitry 150 may not be a component of the console device 140, but may be realized by a separate processing device capable of communicating with the console device 140. The processing device may be, for example, a workstation connected to the X-ray CT apparatus 100 or a cloud server capable of collectively executing processing equivalent to that of the processing circuitry 150.

[0045] The system control function 151 controls various functions of the processing circuit 150 based on input operations received by the input interface 143 .

[0046] The imaging condition determination function 152 determines imaging conditions for medical images by using the trained model MDL defined by the model information in the memory 141. The imaging conditions include, for example, at least one of an imaging protocol for imaging the subject P, preprocessing conditions for detection data or projection data, and reconstruction conditions for reconstructing a CT image.

[0047] The imaging condition determination function 152 may determine the imaging conditions for the medical image in response to an operation input to the input interface 143 .

[0048] The pre-processing function 153 performs pre-processing such as logarithmic conversion processing, offset correction processing, inter-channel sensitivity correction processing, and beam hardening correction on the detection data output by the DAS 116 to generate projection data and store the generated projection data in the memory 141.

[0049] The reconstruction processing function 154 performs reconstruction processing using a filtered back projection method, an iterative reconstruction method, or the like on the projection data generated by the preprocessing function 153 to generate a CT image, which is one type of medical image, and stores the generated CT image in the memory 141.

[0050] The image processing function 155 converts the CT image into a three-dimensional image or cross-sectional image data of an arbitrary cross section by a known method based on the input operation received by the input interface 143. The conversion into a three-dimensional image may be performed by the pre-processing function 153.

[0051] The output control function 156 displays on the display 142 CT images generated through reconstruction processing by the reconstruction processing function 154 and images (e.g., three-dimensional images and cross-sectional images) generated by the image processing function 155. The output control function 156 may also transmit these images to the learning device 200 or other external devices via the communication interface 144.

[0052] In addition, the output control function 156 uses the learned model MDL defined by the model information in the memory 141 to display the medical examination results of the subject P on the display 142.

[0053] [Learning device configuration] 3 is a diagram illustrating an example of the configuration of a learning device 200 according to an embodiment. The learning device 200 includes, for example, a communication interface 202, an input interface 204, a display 206, a memory 208, and a processing circuit 210.

[0054] The communication interface 202 includes, for example, a NIC, a wireless communication module, etc. The communication interface 202 communicates with the X-ray CT apparatus 100 and other external devices via the communication network NW.

[0055] The input interface 204 accepts various input operations by an operator and outputs an electrical signal indicating the content of the accepted input operation to the processing circuit 210. For example, the input interface 204 may be realized by a mouse, keyboard, touch panel, drag ball, switch, button, joystick, foot pedal, camera, infrared sensor, microphone, etc. Note that in this specification, the input interface 204 is not limited to an interface having physical operation components such as a mouse and keyboard. For example, 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 this electrical signal to a control circuit is also included as an example of the input interface 204.

[0056] The display 206 displays various types of information. For example, the display 206 displays the processing results of the processing circuit 210, a GUI that accepts various operations by an operator, etc. The display 206 is, for example, an LCD or an organic EL display.

[0057] The memory 208 is realized by, for example, a semiconductor memory element such as RAM or flash memory, a hard disk, or an optical disk. These storage media may also be realized by other storage devices connected via a communication network NW, such as a NAS or an external storage server device. The memory 208 may also include storage media such as ROM and registers.

[0058] The processing circuit 210 includes, for example, an acquisition function 212, a learning function 214, and an output control function 216. The processing circuit 210 realizes these functions by, for example, a hardware processor (computer) executing a program stored in the memory 208.

[0059] The hardware processor refers to a circuit such as a CPU, GPU, application-specific integrated circuit, or programmable logic device (e.g., a simple programmable logic device, a composite programmable logic device, or a field programmable gate array). Instead of storing a program in memory 208, the program may be directly embedded in the circuit of the hardware processor. In this case, the hardware processor realizes its functions by reading and executing the program embedded in the circuit. The program may be pre-stored in memory 208, or may be stored on a non-transitory storage medium such as a DVD or CD-ROM. The non-transitory storage medium may be inserted into a drive (not shown) of learning device 200 and installed from the non-transitory storage medium into memory 208. The hardware processor is not limited to a single circuit; it may be configured as a single hardware processor by combining multiple independent circuits to realize each function. Furthermore, multiple components may be integrated into a single hardware processor to realize each function.

[0060] The acquisition function 212 acquires a CT image from the X-ray CT device 100 via the communication interface 202. A medical result (treated as a correct medical result) previously determined by a doctor or the like is associated with this CT image as a label. The medical result associated with the CT image is also called a target. In other words, the acquisition function 212 acquires training data in which the correct medical result is associated with the CT image.

[0061] The learning function 214 uses the training data acquired by the acquisition function 212 to learn the unlearned model MDL.

[0062] The output control function 216 transmits the model MDL learned by the learning function 214, that is, model information defining the learned model MDL, to the X-ray CT apparatus 100 via the communication interface 202.

[0063] [Overall flow of medical image diagnostic system (training)] The following describes a series of processes performed by the medical image diagnostic system 1 in this embodiment. Fig. 4 is a flowchart showing the flow of a series of processes performed during training by the medical image diagnostic system 1 in this embodiment.

[0064] First, the imaging condition determination function 152 of the X-ray CT apparatus 100 determines the imaging conditions for a CT image of the subject P (step S100).

[0065] For example, the imaging condition determination function 152 may determine an imaging protocol for the subject P according to settings input to the input interface 143, that is, settings input to the input interface 143 by an imaging staff member or the like.

[0066] Next, the system control function 151 of the X-ray CT apparatus 100 scans the subject P by controlling various functions of the processing circuitry 150 based on the imaging conditions such as the imaging protocol determined by the imaging condition determination function 152 (step S102).

[0067] Next, the pre-processing function 153 and the reconstruction processing function 154 of the X-ray CT apparatus 100 generate a CT image of the subject P (step S104). For example, the pre-processing function 153 generates projection data based on the detection data output by the DAS 116. The reconstruction processing function 154 performs reconstruction processing on the projection data generated by the pre-processing function 153 to generate a CT image of the subject P. The CT image generated in S104 may be a two-dimensional image or a three-dimensional image.

[0068] The system control function 151 acquires the medical treatment results for the CT image of the subject P via the input interface 143 (step S106). For example, when the system control function 151 generates a CT image of the subject P, it instructs the output control function 156 to output the CT image. In response to this, the output control function 156 displays the CT image on the display 142. For example, a doctor looks at the CT image on the display 142, determines whether the subject P has a disease, and inputs this information into the input interface 143. The system control function 151 acquires the information input to the input interface 143 as the medical treatment results. Then, the system control function 151 stores a data set combining the CT image of the subject P and the medical treatment results for the subject P in the memory 141 as training data.

[0069] Next, the acquisition function 212 of the learning device 200 acquires teacher data from one or more X-ray CT devices 100 via the communication interface 202 (step S108).

[0070] Next, the learning function 214 of the learning device 200 aligns the positions of the multiple CT images acquired as training data by the acquisition function 212 (step S110).

[0071] Next, the learning function 214 learns the unlearned model MDL using the training data (step S112).

[0072] For example, the learning function 214 inputs a CT image into an untrained model MDL, and adjusts the parameters (weighting coefficients, bias components, etc.) of the model MDL so that the medical outcome output by the model MDL approaches the correct medical outcome (target medical outcome). The medical outcome may be expressed, for example, as a vector indicating the presence or absence of a specific disease, or a vector indicating its probability (likelihood). The learning function 214 adjusts the parameters of the model MDL using SGD (Stochastic Gradient Descent), Momentum SGD, AdaGrad, RMSprop, AdaDelta, Adam (Adaptive moment estimation), etc., so that the difference between the medical outcome vector output by the model MDL and the target medical outcome vector becomes smaller.

[0073] Next, the learning function 214 inputs the CT image that has been aligned with the learned model MDL, and acquires a contribution map from the learned model MDL (step S114).

[0074] For example, if the CT image is a two-dimensional image, the contribution map may be two-dimensional data, and if the CT image is a three-dimensional image, the contribution map may be three-dimensional data. Furthermore, the contribution map may be time-series data in which the contribution is associated with each position on the CT image in time series.

[0075] In response to this, the output control function 216 transmits the contribution map and model information of the learned model MDL to the X-ray CT apparatus 100 via the communication interface 202. This completes the processing of this flowchart (training processing).

[0076] FIG. 5 is a diagram illustrating a method for acquiring a contribution map during training. For example, assume that CT images of three subjects P1, P2, and P3 are obtained. In this case, the learning function 214 aligns the three CT images so that certain characteristic parts coincide when the three CT images are superimposed. For example, assume that all CT images are images obtained by scanning the upper body. In this case, the learning function 214 may align the three CT images so that characteristic parts, such as the scapulae, are in the same position.

[0077] After aligning multiple CT images, the learning function 214 inputs a composite image of the multiple CT images into the trained model MDL, or inputs a representative CT image from among the multiple CT images into the trained model MDL. As described above, when a CT image is input, the trained model MDL outputs a contribution map and a medical result. This contribution map is used to determine imaging conditions (such as imaging protocols) at runtime, as described below. In the contribution map shown in the figure, dark areas (close to black) represent areas with high contribution, and light areas (close to white) represent areas with low contribution.

[0078] [Overall flow of medical image diagnostic system (runtime)] A series of processes of the medical image diagnostic system 1 in the embodiment will be described below. Fig. 6 is a flowchart showing the flow of a series of processes at runtime by the medical image diagnostic system 1 in the embodiment. Runtime here means executing a process of estimating a medical outcome using the trained model MDL.

[0079] First, the system control function 151 of the X-ray CT apparatus 100 controls various functions of the processing circuitry 150 based on arbitrary imaging conditions to scan the subject P' to be examined (step S200). The subject P' scanned at runtime may be the same as the subject P scanned during training, or may be different. The scan in S200 may be, for example, scanography performed before the actual scan.

[0080] Next, the pre-processing function 153 and the reconstruction processing function 154 of the X-ray CT device 100 generate a CT image of the subject P' (step S202). For example, the pre-processing function 153 generates projection data based on the detection data output by the DAS 116. The reconstruction processing function 154 generates a CT image of the subject P' by performing reconstruction processing on the projection data generated by the pre-processing function 153. The CT image generated in S202 is used for alignment with the contribution map and is not intended to be input into the trained model MDL to obtain a medical result. Therefore, even if the CT image input into the trained model MDL is a three-dimensional image, a two-dimensional CT image (two-dimensional projected still image) may be generated in the processing of S202.

[0081] Next, the imaging condition determination function 152 of the X-ray CT apparatus 100 aligns the contribution map with the CT image of the subject P' based on the contribution map transmitted by the learning device 200 during training (step S204).

[0082] For example, the imaging condition determination function 152 performs image processing such as enlarging, reducing, rotating, and shifting on the contribution map so that characteristic parts on the contribution map match characteristic parts on the CT image of the subject P'. This aligns the contribution map with the CT image of the subject P'. Hereinafter, the contribution map aligned with the CT image of the subject P' will be referred to as a "position-corrected contribution map."

[0083] Next, the imaging condition determination function 152 determines new imaging conditions for the CT image of the subject P' based on the position-corrected contribution map (step S206).

[0084] For example, the imaging condition determination function 152 may determine an imaging protocol that reduces the X-ray dose in areas with low contribution and increases the X-ray dose in areas with high contribution, thereby improving the quality (SN ratio) of the CT image of the subject P' while reducing the exposure dose of the subject P'.

[0085] Furthermore, for example, the imaging condition determination function 152 may determine an imaging protocol that shortens the X-ray irradiation time for regions with low contribution and lengthens the X-ray irradiation time for regions with high contribution. The irradiation time may be, for example, the width of the X-ray irradiation pulse or the integral time. This makes it possible to improve the quality (SN ratio) of the CT image of the subject P' while reducing the radiation exposure dose of the subject P'.

[0086] Furthermore, for example, the imaging condition determination function 152 may determine an imaging protocol such that X-rays irradiated to regions with high contribution rates are detected by the high-resolution X-ray detector 115. This improves the spatial resolution of regions with high contribution rates.

[0087] Furthermore, for example, the imaging condition determination function 152 may determine an imaging protocol in which a spatial filter (for example, a median filter) is applied to the boundary between a region with high contribution and a region with low contribution, thereby reducing image noise and improving the quality of the CT image of the subject P'.

[0088] Furthermore, for example, the imaging condition determination function 152 may determine an imaging protocol in which the energy bins in photon counting are coarse for regions with low contribution and fine for regions with high contribution, thereby improving the energy resolution.

[0089] Furthermore, for example, the imaging condition determination function 152 may determine an imaging protocol such that the current or voltage of the X-ray tube 111 is reduced for regions with low contribution and the current or voltage of the X-ray tube 111 is increased for regions with high contribution, thereby improving the quality (SN ratio) of the CT image of the subject P'.

[0090] Next, the system control function 151 controls various functions of the processing circuitry 150 based on the imaging conditions determined by the imaging condition determination function 152, thereby scanning the subject P' again (step S208). The scan in S208 may be a main scan.

[0091] Next, the preprocessing function 153 and the reconstruction processing function 154 generate a CT image of the subject P' (step S210).

[0092] Next, the output control function 156 inputs the CT image of the subject P' generated in S210 into the trained model MDL (step S212).

[0093] Next, the output control function 156 acquires the medical treatment results and contribution map of the subject P' from the trained model MDL to which the CT image of the subject P' has been input (step S214).

[0094] Next, the output control function 156 determines whether or not the scan of the subject P' should be continued by referring to the imaging protocol, etc. (step S216), and if the scan of the subject P' should not be continued, displays the medical results of the subject P' on the display 142 and terminates the processing of this flowchart (runtime processing).

[0095] 7 to 9 are diagrams schematically showing a method for determining imaging conditions using a contribution map. For example, the imaging condition determination function 152 may determine an imaging protocol (imaging condition), which is one of the imaging conditions, using a contribution map obtained during training, as shown in FIG.

[0096] Furthermore, the imaging condition determination function 152 may determine preprocessing conditions (conditions related to various preprocessing such as logarithmic conversion processing, offset correction processing, inter-channel sensitivity correction processing, and beam hardening correction), which are one of the imaging conditions, using a contribution map obtained during training, as shown in Fig. 8. Furthermore, the imaging condition determination function 152 may determine reconstruction conditions, which are one of the imaging conditions, using a contribution map obtained during training, as shown in Fig. 9.

[0097] [Method for aligning contribution maps; Part 1] FIG. 10 is a diagram illustrating a method for aligning contribution maps. For example, suppose a trained model MDL is created based on CT images of an unspecified number of subjects P, such as P1, P2, and P3, and a contribution map is generated using the trained model MDL. On the other hand, the imaging system of the X-ray CT device 100 may have high hardware reproducibility. In such a case, alignment of contribution maps derived from an unspecified number of subjects P is unnecessary. In other words, there is no need to scan the subject P' to align the contribution maps. Therefore, the imaging condition determination function 152 may omit the process of S204 at runtime and determine new imaging conditions for the CT image of the subject P' based on the contribution map transmitted by the learning device 200 during training.

[0098] [Method for aligning contribution maps; Part 2] FIG. 11 is a diagram illustrating a method for aligning the contribution maps. Contrary to the example in FIG. 10, there are cases where the reproducibility of the imaging system of the X-ray CT device 100 is low in terms of hardware. The imaging system includes the X-ray tube 111, the wedge 112, the collimator 113, the X-ray detector 115, the rotating frame 117, and the like. In such cases, the imaging condition determination function 152 aligns the contribution maps at runtime. For example, the imaging condition determination function 152 may align the contribution maps derived from an unspecified number of subjects P, such as P1, P2, and P3, using a two-dimensional CT image obtained by scanography or the first three-dimensional CT image. Then, the imaging condition determination function 152 determines new imaging conditions for the CT image of the subject P′ based on the contribution map aligned with the CT image of the subject P′, i.e., the position-corrected contribution map.

[0099] [Method for aligning contribution maps; Part 3] 12 is a diagram for explaining a method of aligning the contribution maps. As shown in the figure, the imaging condition determination function 152 may determine new imaging conditions for a CT image of the subject P' by combining contribution maps derived from an unspecified number of subjects P, such as subjects P1, P2, and P3, with a contribution map derived from the subject P'. The contribution maps derived from the unspecified number of subjects P are an example of a "first map," and the contribution map derived from the subject P' is an example of a "second map."

[0100] In the runtime flowchart of Fig. 6, if the scan is continued in the process of S216, the process returns to S204. At this time, the imaging condition determination function 152 may combine the contribution maps derived from an unspecified number of subjects P with the contribution map derived from the subject P' (the contribution map obtained in the process of S214), and determine the imaging conditions based on the combined contribution map (an example of a "third map"). The combination ratio of the contribution maps derived from the unspecified number of subjects P to the contribution map derived from the subject P' may be set to a ratio of 1:3, for example, such that the weight of the contribution map derived from the subject P' is greater than that of the contribution map derived from the subject P. This makes it possible to determine imaging conditions that are more suited to the subject P'.

[0101] [Method for aligning contribution maps; Part 4] FIG. 13 is a diagram illustrating a method for aligning the contribution maps. As shown in the figure, the imaging condition determination function 152 may combine past contribution maps derived from the subject P' to determine new imaging conditions for a CT image of the subject P'. For example, assume that contribution maps derived from the subject P' were obtained one year ago, two years ago, and three years ago. In this case, the imaging condition determination function 152 may combine these three contribution maps derived from the subject P' and determine imaging conditions based on the combined contribution map. The combination ratio may be, for example, one year ago:two years ago:three years ago=5:3:1. In other words, the more recent the contribution map, the greater the weight. This makes it possible to determine imaging conditions that are more suitable for the current subject P'.

[0102] [How to determine imaging conditions; Part 1] 14 is a diagram for explaining a method for determining imaging conditions. For example, the imaging condition determination function 152 may determine the imaging angle based on the contribution map. Specifically, as shown in the figure, the imaging condition determination function 152 may determine an imaging protocol such that a larger dose of X-rays is irradiated onto the front of a subject P' lying on a bed, and a smaller dose of X-rays is irradiated onto the side of the subject P'.

[0103] [How to determine imaging conditions; Part 2] 15 is a diagram for explaining a method for determining imaging conditions. For example, on the contribution map, there may be cases where parts with high contribution (dark parts) are present at multiple separated locations. In this case, the imaging condition determination function 152 may determine an imaging protocol such that a high dose of X-rays is irradiated obliquely onto a subject P' lying on a bed, as shown in the figure.

[0104] [How to determine imaging conditions; Part 3] FIG. 16 is a diagram illustrating a method for determining imaging conditions. For example, the imaging condition determination function 152 may change the imaging protocol in accordance with the rotation of the imaging system. Specifically, assume that at a certain sampling time k, X-rays are irradiated from the X-ray tube 111(k) and detected by the X-ray detector 115(k). In this case, the processing circuit 150 generates a CT image based on the X-ray detection data detected by the X-ray detector 115(k), and further inputs the CT image into the trained model MDL to generate a contribution map at time k. The imaging condition determination function 152 determines the imaging protocol for the next time k+1 based on the contribution map at time k. As a result, the dose of X-rays irradiated from the X-ray tube 111(k+1) at time k+1 is determined, and the positions of the X-ray tube 111(k+1) and the X-ray detector 115(k+1) when irradiating X-rays are determined.

[0105] [Method for determining imaging conditions; Part 4] 17 is a diagram for explaining a method for determining imaging conditions. For example, the imaging condition determination function 152 may change the imaging protocol each time the imaging system makes one rotation. Specifically, when the time advances from 1 to k-1 to k and the imaging system makes one rotation at time k, the processing circuit 150 generates a CT image based on the detection data of X-rays detected by the X-ray detector 115(k) at time k, and further inputs the CT image into the trained model MDL to generate a contribution map at time k. Then, the imaging condition determination function 152 determines the imaging protocol for the next cycle k+1 to 2k, in which the imaging system makes one rotation, based on the contribution map at time k.

[0106] [Method for determining imaging conditions; Part 5] 18 is a diagram for explaining a method for determining imaging conditions. The imaging protocol also includes a protocol for reading out detection data from the X-ray detector 115 (hereinafter referred to as a data readout protocol). Therefore, the imaging condition determination function 152 may determine the data readout protocol based on the contribution map. For example, the imaging condition determination function 152 may determine a data readout protocol such that the number of bindings is increased in regions with lower contributions on the contribution map and the number of bindings is reduced in regions with higher contributions. This makes it possible to improve the quality of CT images by improving the spatial resolution in regions with high contributions while reducing noise by lowering the spatial resolution in regions with low contributions.

[0107] [Method for determining imaging conditions; Part 6] FIG. 19 is a diagram illustrating a method for determining imaging conditions. The imaging condition determination function 152 may dynamically change the energy BIN range for photon counting based on a contribution map. For example, assume that the BIN range is changeable in the X-ray detector 115. Furthermore, assume that contribution maps obtained are contribution map 1, which indicates a spatial distribution where low energy is effective, contribution map 2, which indicates a spatial distribution where medium energy is effective, and contribution map 3, which indicates a spatial distribution where high energy is effective. In this case, the imaging condition determination function 152 changes the BIN threshold for each detection region according to contribution maps 1, 2, and 3. For example, in the region on the body axis direction (on the Z-axis) in FIG. 19, low energy is effective and high energy has little effect, so the low energy BIN is made finer and the high energy BIN is made coarser. This allows for fine energy resolution of the region with high contribution.

[0108] [Method for determining imaging conditions; Part 7] 20 is a diagram for explaining a method for determining imaging conditions. The imaging condition determination function 152 may dynamically change the energy when irradiating X-rays depending on the contribution. As shown in the figure, for example, assume that there are contribution maps obtained by irradiating X-rays with an energy of 80 keV and contribution maps obtained by irradiating X-rays with an energy of 120 keV. In this case, the imaging condition determination function 152 may determine an imaging protocol such that the energy is increased to 80 keV for regions with high contribution on the 80 keV contribution map, and the energy is increased to 120 keV for regions with high contribution on the 120 keV contribution map.

[0109] [Method for determining imaging conditions; Part 7] 21 is a diagram for explaining a method for determining imaging conditions. The imaging condition determination function 152 may determine an imaging protocol according to a change in the contribution degree over time. As described above, the contribution degree map may be time-series data in which the contribution degree is associated with each position on the CT image in a time series.

[0110] For example, scan A at time t1 is imaging performed when a contrast agent is administered to a blood vessel, and scan B at time t2 and scan C at time t3 are imaging of the myocardium. The processing circuitry 150 inputs the CT image obtained by scan A into the trained model MDL to generate contribution maps for future times t2 and t3. The imaging condition determination function 152 determines the imaging protocol for the next time t2 and the imaging protocol for the time t3 after that, based on the contribution maps for times t2 and t3.

[0111] Furthermore, the processing circuitry 150 may repeatedly generate a contribution map by inputting the CT image into the trained model MDL each time a scan is repeated, such as by continuous rotation of the imaging system of the X-ray CT device 100. Then, the processing circuitry 150 may terminate the scan when the contribution of the contribution map exceeds a threshold (required contribution).

[0112] [Method for determining imaging conditions; Part 8] 22 and 23 are diagrams for explaining a method for determining imaging conditions. For example, as shown in Fig. 22, a contribution map may be generated so that the contribution also increases around the time when the blood vessel of interest stains most deeply after the start of angiography (around 2 seconds in the figure). In this case, as shown in Fig. 23, the imaging condition determination function 152 may determine an imaging protocol that increases the X-ray dose around the time when the blood vessel stains most deeply (around 2 seconds in the figure).

[0113] [Method for determining imaging conditions; Part 9] 24 and 25 are diagrams for explaining a method for determining imaging conditions. As shown in FIGS. 24 and 25, a contribution map along the body axis direction (Z direction) of the subject P' can also be generated. For example, assume that a scan is started from the head of the subject P' along the body axis direction (Z direction). In this case, the imaging condition determination function 152 may determine an imaging protocol such that the scan is terminated when the scan position reaches a position on the contribution map where the contribution is equal to or less than a threshold value while the scan position gradually shifts from the head to the feet of the subject P'.

[0114] 24, the degree of contribution is uniformly high from the head to the toes of the subject P'. In this case, the imaging condition determination function 152 determines an imaging protocol that scans the entire body of the subject P' from the head to the toes without terminating the scan midway, while changing the X-ray dose etc. in accordance with the change in the degree of contribution.

[0115] 25, the contribution is high only in the vicinity of the chest of the subject P'. In this case, the imaging condition determination function 152 determines an imaging protocol in which the X-ray dose is increased around the chest, and then scanning is terminated from the abdomen onwards (from Z1 onwards in the figure).

[0116] [Method for determining imaging conditions; Part 10] 26 and 27 are diagrams for explaining a method for determining imaging conditions. For example, when scanning a subject P' while the imaging system rotates, such as at k, k+1, k+2, and k+3, a contribution map is generated for each scan position. In this case, the imaging condition determination function 152 may determine an imaging protocol for reducing the radiation exposure dose of the subject P' at each scan position based on the contribution map for each scan position. Specifically, as shown in FIG. 27, in each of the contribution maps k, k+1, k+2, and k+3, a region with low contribution (a region where the contribution is equal to or less than a threshold) is set as an X-ray dose reduction region. In the dose reduction region, for example, a lead collimator that blocks X-rays may be physically installed, or a compensation filter that reduces the X-ray output may be installed.

[0117] [Method for determining imaging conditions; Part 11] 28 is a diagram for explaining a method for determining imaging conditions. As shown in the figure, the imaging condition determination function 152 may set an X-ray dose reduction region according to the contribution of the subject P' in the body axis direction (Z direction). For example, in a neck contribution map, the contribution is large from the center to the bottom of the map. In this case, the neck dose reduction region is set at the right and left ends of the map. In a chest contribution map, the contribution of the entire map is large. In this case, no chest dose reduction region is set. In an abdomen contribution map, only the contribution of the lower right part of the map is large. In this case, the abdomen dose reduction region is set to the entire left side as viewed from the center of the map.

[0118] According to the embodiment described above, when a CT image is input, the learning device 200 learns the model MDL so as to output the medical results and contribution map of the person included in the input CT image. When the X-ray CT device 100 captures an image of the subject P' and generates a CT image, it determines the imaging conditions for the CT image of the subject P' based on the CT image and the contribution map output by the trained model MDL during training. As a result, while imaging and analysis mutually influence each other, it is possible to obtain medical images that enable highly accurate diagnosis.

[0119] Furthermore, according to the above-described embodiment, an imaging protocol is determined that reduces the X-ray dose in areas with low contribution and increases the X-ray dose in areas with high contribution, thereby improving the quality (SN ratio) of the CT image of the subject P' while reducing the exposure dose of the subject P'.

[0120] Furthermore, according to the above-described embodiment, an imaging protocol is determined that shortens the X-ray irradiation time for areas with low contribution and lengthens the X-ray irradiation time for areas with high contribution, thereby improving the quality (SN ratio) of the CT image of the subject P' while reducing the amount of radiation exposure to the subject P'.

[0121] Furthermore, according to the above-described embodiment, an imaging protocol is determined such that X-rays irradiated to areas with high contribution are detected by a high-resolution X-ray detector 115, thereby improving the spatial resolution of areas with high contribution.

[0122] Furthermore, according to the above-described embodiment, an imaging protocol is determined in which a spatial filter (e.g., a median filter) is applied to the boundary between the high-contribution region and the low-contribution region, thereby reducing image noise. As a result, the quality of the CT image of the subject P' can be improved.

[0123] Furthermore, according to the above-described embodiment, an imaging protocol is determined in which the number of BINs for photon counting is reduced for areas with low contribution and the number of BINs is increased for areas with high contribution, thereby improving energy resolution.

[0124] Furthermore, according to the above-described embodiment, an imaging protocol is determined in which the current or voltage of the X-ray tube 111 is reduced for areas with low contribution and the current or voltage of the X-ray tube 111 is increased for areas with high contribution, thereby improving the quality (SN ratio) of the CT image of the subject P'.

[0125] (Other embodiments (modifications)) Modifications of the above-described embodiment will be described below. In the above-described embodiment, the X-ray CT apparatus 100, which is a modality, and the learning apparatus 200, which is a workstation or a cloud server, are described as separate apparatuses, but this is not limiting.

[0126] For example, the learning device 200 may be one function of the X-ray CT device 100. Specifically, the X-ray CT device 100 may have various functions of the learning device 200 (such as the acquisition function 212 and the learning function 214). Furthermore, the X-ray CT device 100 may be one function of the learning device 200. Specifically, the learning device 200 may have various functions of the X-ray CT device 100 (such as the imaging condition determination function 152). In this way, the X-ray CT device 100 and the learning device 200 may be an integrated device. A medical image diagnostic system 1 equipped with these devices is another example of a "medical image diagnostic device."

[0127] Although the above-described embodiment has been described assuming that there is only one trained model MDL, this is not limiting. For example, if there are multiple specific diseases to be diagnosed, a trained model MDL may be generated for each disease. For example, assume that there is a trained model MDL1 for lung cancer, a trained model MDL2 for pleurisy, and a trained model MDL3 for bronchial asthma. In this case, the same CT image is input to each of the three trained models MDL1 to MDL3. Each of the three trained models MDL1 to MDL3 outputs a contribution map. Different diseases may require different regions of interest on the CT image, and the contribution maps may also differ accordingly. In such cases, the contribution maps for each trained model MDL may be combined into a single contribution map by adding or averaging them.

[0128] In the above-described embodiment, the model MDL is trained based on training data in which correct medical results are associated with the CT image as labels (targets) to output a contribution map and a medical result when a CT image is input. However, this is not limited to this. For example, the model MDL may be a model combining a model A that has been sufficiently trained in advance and a model B that has also been sufficiently trained in advance. Model A may be trained to output a contribution map when a CT image is input, for example, a DNN that functions as an attention mechanism. Model B may be trained to output a medical result when a CT image is input, for example, a CNN. In this way, multiple models pre-trained for each function, such as a function to estimate a medical result and a function to generate a contribution map, may be combined into one to form the "model MDL" according to this embodiment. In this case, the model MDL does not need to be trained based on training data.

[0129] Although several embodiments have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their modifications are included within the scope and spirit of the invention, as well as within the scope of the invention and its equivalents as defined in the claims. [Explanation of symbols]

[0130] 1... medical image diagnostic system, 100... X-ray CT device, 110... stand device, 130... bed device, 140... console device, 141... memory, 142... display, 143... input interface, 144... communication interface, 150... processing circuit, 151... system control function, imaging condition determination function, 153... pre-processing function, 154... reconstruction processing function, 155... image processing function, 156... output control function, 200... learning device, 202... communication interface, 204... input interface, 206... display, 208... memory, 210... processing circuit, 212... acquisition function, 214... learning function, 216... output control function

Claims

1. an acquisition unit for acquiring a medical image of a subject; a determination unit that inputs the medical image of the subject acquired by the acquisition unit into a machine learning model that is trained based on teacher data in which medical images are associated with medical results of people included in the medical images, and determines imaging conditions for the medical image of the subject based on the medical results of the subject output by the machine learning model in response to the input of the medical image of the subject and a map in which each position on the medical image is associated with a contribution to the medical results, The contribution degree is an index indicating the relative extent to which the machine learning model paid attention to each position on the medical image during the process from when the medical image was input to when the medical result was output. Medical imaging diagnostic equipment.

2. The machine learning model is a neural network including a convolutional layer and an attention mechanism that calculates the contribution. The medical image diagnostic apparatus according to claim 1 .

3. The attention mechanism calculates the degree of attention for each pixel as the contribution by multiplying each pixel on the feature map converted from the medical image by a different weight. The medical image diagnostic apparatus according to claim 2 .

4. the determination unit determines the imaging conditions based on the map output by the machine learning model during learning and the medical image of the subject acquired by the acquisition unit. The medical image diagnostic apparatus according to claim 1 .

5. The determination unit aligning the map output by the machine learning model during learning with the medical image of the subject acquired by the acquisition unit; determining the imaging conditions based on the map registered with the medical image of the subject; The medical image diagnostic apparatus according to claim 4.

6. the acquisition unit newly acquires a medical image of the subject captured based on the imaging conditions; the determination unit inputs medical images of the subject previously acquired by the acquisition unit into the machine learning model, and newly determines the imaging conditions based on a first map that is the map output by the machine learning model during learning, a second map that is the map output by the machine learning model to which the medical images have been input, and medical images of the subject newly acquired by the acquisition unit; 6. The medical image diagnostic apparatus according to claim 4 or 5.

7. The determination unit weighting and combining the first map and the second map; performing registration between the synthesized third map and the medical image of the subject newly acquired by the acquisition unit; determining the imaging conditions based on the third map that has been registered with the medical image of the subject; The medical image diagnostic apparatus according to claim 6.

8. the determination unit increases the weight of the second map compared to the weight of the first map. The medical image diagnostic apparatus according to claim 7.

9. the acquisition unit acquires a new medical image of the subject each time the imaging conditions are determined, the determination unit inputs the medical image of the subject to the machine learning model every time the medical image of the subject is acquired, and weights and combines the first map and the second maps output by the machine learning model every time the medical image of the subject is input.

9. The medical image diagnostic apparatus according to claim 7 or 8.

10. the determination unit increases the weight of a second map outputted by the machine learning model more recently among the plurality of second maps; The medical image diagnostic apparatus according to claim 9.

11. The imaging conditions include at least one of an imaging protocol for imaging the subject, a preprocessing condition for data obtained by scanning the subject, and a reconstruction condition for the medical image from the data. The medical image diagnostic apparatus according to any one of claims 1 to 10.

12. The map is time-series data in which the contribution degree is associated with each position on the input medical image in a time series manner. The medical image diagnostic apparatus according to any one of claims 1 to 11.

13. the medical image is a medical image taken by X-ray, the determination unit determines, as the imaging condition, an imaging protocol that reduces a dose of the X-rays irradiated to an area on the map having a low degree of contribution. The medical image diagnostic apparatus according to any one of claims 1 to 12.

14. the medical image is a medical image taken by X-ray, the determination unit determines, as the imaging condition, an imaging protocol that increases the dose of the X-rays irradiated to the region on the map that has a high degree of contribution. The medical image diagnostic apparatus according to any one of claims 1 to 13.

15. the medical image is a medical image taken by X-ray, the determination unit determines, as the imaging condition, an imaging protocol that shortens the irradiation time of the X-rays to the region on the map that has a low degree of contribution. The medical image diagnostic apparatus according to any one of claims 1 to 12.

16. the medical image is a medical image taken by X-ray, the determination unit determines, as the imaging condition, an imaging protocol that lengthens the irradiation time of the X-rays to the region on the map where the contribution is high. The medical image diagnostic apparatus according to any one of claims 1 to 15.

17. the medical image is a medical image taken by X-ray, the determination unit determines, as the imaging condition, an imaging protocol that reduces the number of binding photons in the region on the map where the contribution is low.

17. A medical image diagnostic apparatus according to any one of claims 1 to 16.

18. the medical image is a medical image taken by X-ray, the determination unit determines, as the imaging condition, an imaging protocol that increases the number of binding photons in the region on the map where the contribution is high.

18. A medical image diagnostic apparatus according to any one of claims 1 to 17.

19. the medical image is a medical image taken by X-ray, the determination unit determines, as the imaging condition, an imaging protocol that reduces a current or a voltage of an X-ray tube that irradiates the X-rays in an area on the map where the degree of contribution is low.

19. A medical image diagnostic apparatus according to any one of claims 1 to 18.

20. the medical image is a medical image taken by X-ray, the determination unit determines, as the imaging condition, an imaging protocol for increasing a current or a voltage of an X-ray tube that irradiates the X-rays in an area on the map where the degree of contribution is high.

20. A medical image diagnostic apparatus according to any one of claims 1 to 19.

21. The determination unit When a plurality of the machine learning models are present, the medical image of the subject is input to each of the plurality of machine learning models, and the maps output by each of the plurality of machine learning models are synthesized; determining the imaging conditions based on the synthesized map; 21. A medical image diagnostic apparatus according to any one of claims 1 to 20.

22. The computer acquiring a medical image of the subject; inputting the acquired medical image of the subject into a machine learning model trained based on training data in which medical images are associated with medical results of people included in the medical images; determining imaging conditions for the medical image of the subject based on the medical result of the subject output by the machine learning model in response to the input of the medical image of the subject and a map in which each position on the medical image is associated with a contribution to the medical result; The contribution degree is an index indicating the relative extent to which the machine learning model paid attention to each position on the medical image during the process from when the medical image was input to when the medical result was output. A method for determining medical imaging conditions.

23. A program to be executed by a computer, acquiring a medical image of a subject; inputting the acquired medical image of the subject into a machine learning model trained based on training data in which the medical image is associated with the medical results of the person included in the medical image; determining imaging conditions for the medical image of the subject based on a medical result of the subject output by the machine learning model in response to input of the medical image of the subject, and a map in which each position on the medical image is associated with a contribution to the medical result; The contribution degree is an index indicating the relative extent to which the machine learning model paid attention to each position on the medical image during the process from when the medical image was input to when the medical result was output. program.

Images