Medical image processing device, medical image processing system, and program

The medical image processing device converts medical data from various modality devices into a compatible format using image conversion models, addressing the challenge of applying images from different devices to a trained model without additional training, thereby maintaining analysis accuracy and simplifying implementation.

JP2024164649APending Publication Date: 2024-11-27CANON MEDICAL SYST CORP

Patent Information

Application Number
JP2023080286
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-15
Publication Date
2024-11-27

AI Technical Summary

Technical Problem

The challenge is to apply medical images captured by different modality devices to a trained model without performing additional training, as differences in characteristics such as noise and contrast can lead to decreased analysis accuracy, and additional training is complex and difficult to implement in facilities like hospitals.

Method used

A medical image processing device that converts medical data from various modality devices into a compatible input format using image conversion models and conditions, allowing images to be applied to a trained model without additional learning, by utilizing image conversion conditions and models to adjust for differences in imaging characteristics.

Benefits of technology

Enables the application of images from different modality devices to a trained model, maintaining analysis accuracy while avoiding the complexity and constraints of additional training, thus ensuring consistent performance across diverse imaging equipment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024164649000001_ABST
    Figure 2024164649000001_ABST
Patent Text Reader

Abstract

To enable images captured by different modality devices to be applied to a learned model without additional learning for the learned model.SOLUTION: A medical image processing device according to an embodiment includes: an acquisition unit that acquires medical data; and a conversion unit that converts the medical data acquired by the acquisition unit into an input image conforming to an input of a learned model that outputs an analysis result of an image.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The embodiments disclosed in this specification and the drawings relate to a medical image processing apparatus, a medical image processing system, and a program. [Background technology]

[0002] Conventionally, trained models have been generated for analyzing medical images captured by various modality devices such as X-ray CT (Computed Tomography) devices and MRI (Magnetic Resonance Imaging) devices. To prevent a decrease in the analysis accuracy of such trained models, it is necessary to input optimal images to the trained models.

[0003] However, when a modality device that captures a medical image used to generate a trained model and a modality device that captures a medical image input to the trained model are modality devices with different manufacturers, performance, etc., the characteristics of the medical image input to the trained model are different from the characteristics of the medical image used to generate the trained model. Then, when a medical image with different characteristics from the medical image used to generate such a trained model is input to the trained model, the accuracy of the output of the trained model may decrease.

[0004] In addition, in order to guarantee the accuracy of the output of the trained model, it is not possible to perform additional training of the trained model in a facility such as a hospital. In other words, it is not possible to obtain medical images stored in a facility such as a hospital, particularly medical images captured by a modality device made by a different manufacturer, and train the trained model. Therefore, it is desirable to be able to apply images captured by a different modality device to the trained model without performing additional training on the trained model. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] International Publication No. 2023 / 281738 [Patent Document 2] Patent Publication No. 2022-35508 [Patent Document 3] JP 2020-39490 A Summary of the Invention [Problem to be solved by the invention]

[0006] One of the problems to be solved by the embodiments disclosed in this specification and the drawings is to make it possible to apply images captured by a different modality device to a trained model without performing additional training on the trained model. However, the problems to be solved by the embodiments disclosed in this specification and the drawings are not limited to the above problems. Problems corresponding to the effects of each configuration shown in the embodiments described later can also be positioned as other problems. [Means for solving the problem]

[0007] A medical image processing device according to an embodiment includes an acquisition unit that acquires medical data, and a conversion unit that converts the medical data acquired by the acquisition unit into an input image that conforms to the input of a trained model that outputs image analysis results. [Brief description of the drawings]

[0008] [Figure 1] 1 is a block diagram showing an example of the arrangement of a medical image processing system according to a first embodiment. [Diagram 2] FIG. 11 is a diagram showing an example of processing until a trained model executes image analysis processing in the medical image processing apparatus according to the first embodiment. [Diagram 3] 1 is a block diagram showing an example of the arrangement of a medical image processing apparatus according to a first embodiment. [Figure 4] FIG. 2 is a diagram showing an example of a first correspondence table in which imaging conditions of raw data collected in a first modality device correspond to image conversion conditions in the medical image processing system according to the first embodiment. [Diagram 5] FIG. 2 is a diagram showing an example of a second correspondence table in which the imaging conditions of a first medical image captured by a first modality device correspond to image conversion method information in the medical image processing system according to the first embodiment. [Figure 6] 4 is a diagram showing an example of a case where an image transformation model is applied to a first medical image in the medical image processing apparatus according to the first embodiment. [Figure 7] FIG. 4 is an explanatory diagram for explaining an example of a calculation method of imaging conditions in a first modality device for a second medical image in the medical image processing apparatus according to the first embodiment. [Figure 8] FIG. 4 is an explanatory diagram for explaining an example of a calculation method of imaging conditions in a first modality device for a second medical image in the medical image processing apparatus according to the first embodiment. [Figure 9] FIG. 11 is an explanatory diagram for explaining another example of the calculation method of the imaging conditions in the first modality device for the second medical image in the medical image processing apparatus according to the first embodiment. [Figure 10] FIG. 11 is an explanatory diagram for explaining another example of the calculation method of the imaging conditions in the first modality device for the second medical image in the medical image processing apparatus according to the first embodiment. [Figure 11] FIG. 4 is a flowchart illustrating the contents of conversion processing executed by the medical image processing apparatus according to the first embodiment. [Figure 12] FIG. 4 is a flowchart illustrating the contents of conversion processing executed by the medical image processing apparatus according to the first embodiment. [Figure 13] 4A to 4C are views showing an example of an output mode of an input image and an operation area outputted to a display of the medical image processing apparatus according to the first embodiment. [Figure 14] 4A to 4C are views showing an example of a selection operation of an image transformation model in an operation area in the medical image processing apparatus according to the first embodiment. [Figure 15] 4A to 4C are views showing an example of a selection operation of an image transformation model in an operation area in the medical image processing apparatus according to the first embodiment. [Figure 16] 5A to 5C are views showing examples of analysis results, image conversion method information, and usage ratio information outputted to a display of the medical image processing apparatus according to the first embodiment. [Figure 17] FIG. 4 is a flowchart illustrating the contents of update processing executed by the medical image processing apparatus of the medical image processing system according to the first embodiment. [Figure 18] FIG. 11 is a block diagram showing an example of the arrangement of a medical image processing apparatus according to a second embodiment. [Figure 19] FIG. 11 is a flowchart illustrating the contents of conversion processing executed by the medical image processing apparatus according to the second embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] Hereinafter, embodiments of a medical image processing device, a medical image processing system, and a program will be described with reference to the drawings. In the following description, components having substantially the same functions and configurations are denoted by the same reference numerals, and duplicated descriptions will be given only when necessary.

[0010] [First embodiment] FIG. 1 is a block diagram showing a configuration example of a medical image processing system according to the first embodiment. The medical image processing system 1 shown in FIG. 1 is a system that allows medical data generated in a modality device to be input to a trained model trained using medical images generated in another modality device. As shown in FIG. 1, the medical image processing system 1 according to this embodiment is configured to include a first modality device 10, a second modality device 20, a third modality device 30, a medical data server 40, and a medical image processing device 50. In the example shown in FIG. 1, the first modality device 10, the second modality device 20, the third modality device 30, the medical data server 40, and the medical image processing device 50 are connected to each other via a network NW so as to be able to communicate with each other.

[0011] The first modality device 10, the second modality device 20, and the third modality device 30 are devices that capture images of a subject or a phantom to generate medical data. The first modality device 10, the second modality device 20, and the third modality device 30 transmit the generated medical data to the medical data server 40 and the medical image processing device 50 via the network NW. The medical data includes raw data collected by capturing images of a subject or a phantom, and medical images generated by capturing images of a subject or a phantom. The medical images are generated by performing a reconstruction process on the raw data. In other words, the raw data is data before the reconstruction process is performed. In the following description, the medical image captured by the first modality device 10 is referred to as the first medical image, the medical image captured by the second modality device 20 is referred to as the second medical image, and the medical image captured by the third modality device is referred to as the third medical image.

[0012] The first modality device 10, the second modality device 20, and the third modality device 30 according to the present embodiment are, for example, X-ray CT devices. Note that the first modality device 10, the second modality device 20, and the third modality device 30 according to the present embodiment are not limited to X-ray CT devices, and may be X-ray diagnostic devices other than X-ray CT devices, or may be an ultrasound diagnostic device, an MRI device, or the like.

[0013] Furthermore, the first modality device 10, the second modality device 20, and the third modality device 30 are modality devices of the same type, such as an X-ray CT device, but are modality devices with different manufacturers, models, performance, installed facilities, etc. In this embodiment, for example, the first modality device 10 and the third modality device 30 are modality devices of the same manufacturer, but are modality devices with different performance, models, installed facilities, etc. In addition, the second modality device 20 and the third modality device 30 are modality devices of different manufacturers. Note that the first modality device 10 and the second modality device 20 are modality devices of different manufacturers.

[0014] In the medical image processing system 1 according to the present embodiment, it is not necessary to provide both the first modality device 10 and the second modality device 20, and it is sufficient to provide either the first modality device 10 or the second modality device 20. In addition, the third modality device 30 may be, for example, a device that is installed in place of the first modality device 10 or the second modality device 20 in a medical field. In this way, when the third modality device 30 is installed in place of the first modality device 10 or the second modality device 20, the first modality device 10 and the second modality device 20 may not be provided in the example shown in FIG. 1. Even if the first modality device 10 and the second modality device 20 are not provided, raw data previously collected by the first modality device 10, the first medical image taken by the first modality device 10, and the second medical image taken by the second modality device 20 are stored in the medical data server 40 described later.

[0015] The medical data server 40 is a server that stores medical data and the like generated by the first modality device 10, the second modality device 20, and the third modality device 30. The medical data server 40 also transmits the stored medical data and the like to the first modality device 10, the second modality device 20, and the third modality device 30, and to the medical image processing device 50, via the network NW. The medical data server 40 according to this embodiment is, for example, an image server such as a PACS (Picture Archiving and Communication System).

[0016] The medical image processing device 50 is a device that performs image conversion processing on medical data. This medical image processing device 50 acquires medical data from the first modality device 10, the second modality device 20, the third modality device 30, the medical data server 40, etc. via the network NW. For this reason, the medical image processing device 50 is, for example, a workstation and an image interpretation terminal, and is provided on a system connected via the network NW. In addition, the medical image processing device 50 according to this embodiment executes image analysis processing on an image input to the trained model using the trained model. Note that the medical image processing device 50 may be an offline device. In that case, the medical image processing device 50 may acquire medical data from the first modality device 10, the second modality device 20, the third modality device 30, the medical data server 40, etc. via a portable recording medium.

[0017] The network NW refers to a general information and communication network that uses electric communication technology. The network NW includes, for example, wireless / wired LANs such as hospital backbone LANs (Local Area Networks) and the Internet, as well as electric communication circuit networks, optical fiber communication networks, cable communication networks, and satellite communication networks.

[0018] Next, an example of processing executed by the medical image processing device 50 according to this embodiment will be described with reference to FIG. 2. FIG. 2 is a diagram showing an example of processing until the trained model executes image analysis processing in the medical image processing device 50 according to this embodiment. In the example shown in FIG. 2, a case will be described in which medical data MD captured by the first modality device 10 and medical data MD captured by the second modality device 20 are input to the trained model 210 trained using the third medical image generated by the third modality device 30. Here, the trained model 210 is a model that executes image analysis such as segmentation and detection on the input image IM. This trained model 210 is generated by machine learning such as reinforcement learning, supervised learning, unsupervised learning, and deep learning.

[0019] The learning method of the trained model 210 is not limited to these learning methods. That is, the learning method of the trained model 210 is arbitrary, and the trained model 210 may be generated by other learning methods. In addition, in this embodiment, the trained model 210 is trained using medical images captured by one modality device of the third modality device 30, but the number of modality devices that capture medical images used to generate the trained model 210 is not limited to one. That is, the number of modality devices that capture medical images used to generate the trained model 210 is arbitrary, and the trained model 210 may be trained using medical images captured by multiple modality devices including the third modality device 30.

[0020] As described above, in this embodiment, the first modality device 10 and the third modality device 30 are modality devices made by the same manufacturer, but are different in model, performance, installed facilities, etc., and the second modality device 20 and the third modality device 30 are modality devices made by different manufacturers. Therefore, the third medical image is different from the first medical image or the second medical image in terms of characteristics of the medical image. Here, the characteristics of the medical image are noise contained in the medical image, contrast of the medical image, image quality, CT value, etc.

[0021] In this way, when the medical image characteristics such as noise and contrast differ between the third medical image and the first or second medical image, even if the first or second medical image is input into the trained model 210 trained using the third medical image, the analysis results of the first or second medical image may not have the intended analysis accuracy.

[0022] In response to this, it is possible to suppress the deterioration of analysis accuracy by additionally training the trained model 210, which has been trained using the third medical image, using the first medical image and / or the second medical image. However, when taking personal information outside as data for training in order to perform additional training of the trained model 210, the user, etc. must carry out various procedures for taking personal information outside and measures against information leakage, and these procedures and measures are cumbersome for the user. In addition, in a facility such as a hospital, it is difficult to perform additional training of the trained model 210 using the first medical image and / or the second medical image from the viewpoint of guaranteeing analysis accuracy.

[0023] Therefore, as shown in FIG. 2, the medical image processing device 50 image-converts the medical data MD generated in the first modality device 10 or the second modality device 20 into an input image IM that matches the trained model 210 trained using the third medical image, and inputs the image-converted input image IM to the trained model 210. Then, the medical image processing device 50 analyzes the input image IM using the trained model 210, and outputs an analysis result AR of the input image IM by the trained model 210. In this way, by image-converting the medical data MD generated in the first modality device 10 or the second modality device 20 into an input image IM that matches the trained model 210 trained using the third medical image, it is possible to apply an image captured by a different modality device to the trained model 210 while suppressing a decrease in analysis accuracy.

[0024] Fig. 3 is a block diagram showing an example of the configuration of a medical image processing device 50 according to this embodiment. As shown in Fig. 3, the medical image processing device 50 is configured to include a memory 51, a display 52, an input interface 53, a communication interface 54, and a processing circuit 55. A plurality of elements constituting the medical image processing device 50 may be stored in one housing, or may be stored in a distributed manner in a plurality of housings.

[0025] The memory 51 is realized by, for example, a semiconductor memory element such as a RAM (Random Access Memory), a flash memory, a hard disk, an optical disk, etc. In this embodiment, the memory 51 stores, for example, medical data MD acquired from the first modality device 10, the second modality device 20, the third modality device, or the medical data server 40, stores the trained model 210, stores image conversion conditions, which are conditions for converting raw data collected in the first modality device 10 into an input image IM, in association with the imaging conditions of the raw data collected in the first modality device, stores an image conversion method, which is a method for converting a first medical image into an input image IM, in association with the imaging conditions of the first medical image, stores image conversion method information, and stores the input image IM. This memory 51 corresponds to the first storage unit and the second storage unit in this embodiment.

[0026] Here, the imaging conditions are conditions used for imaging when generating raw data, the first medical image, and the second medical image. For example, in the case of an X-ray CT device, the imaging conditions include reconstruction conditions such as tube voltage, tube current, and reconstruction function, as well as conditions such as a filter, an imaging region, and a protocol name. Note that the imaging conditions do not necessarily include the imaging region and the protocol name. The imaging conditions may also include conditions other than reconstruction conditions such as tube voltage (kV), tube current, and reconstruction function (FC), as well as a filter, an imaging region, and a protocol name.

[0027] The image conversion conditions are conditions for converting raw data collected in the first modality device 10 into an input image IM. The image conversion conditions are created, for example, based on raw data collected in the first modality device 10, the imaging conditions of the raw data when the raw data is collected in the first modality device 10, and the input image IM. The image conversion conditions are, for example, reconstruction conditions for reconstructing raw data collected in the first modality device 10, and conversion formulas for converting the reconstruction conditions when collecting raw data in the first modality device into reconstruction conditions for reconstructing the input image IM. In this embodiment, since the first modality device 10 and the third modality device 30 are devices from the same manufacturer, it is possible to prepare image conversion conditions for each imaging condition of the raw data collected in the first modality device 10 in advance, and the raw data collected in the first modality device 10 can be converted into the input image IM using the image conversion conditions prepared in advance.

[0028] The image conversion method is a method of converting a first medical image captured in the first modality device 10 into an input image IM. The image conversion method is, for example, a rule-based conversion formula or an image conversion model. The rule-based conversion formula is generated, for example, based on the first medical image and the imaging conditions of the first medical image. The image conversion model is generated by machine learning based on the first medical image and the imaging conditions of the first medical image. More specifically, the image conversion model is generated by machine learning based on a plurality of first medical images including a first medical image captured under optimal conditions for the input image IM, in which the same object (e.g., a phantom) is captured under different imaging conditions in the first modality device 10, and the imaging conditions of the plurality of first medical images including optimal conditions for the input image IM.

[0029] Although the above-mentioned image conversion model is generated by machine learning based on the first medical image captured by the first modality device 10 and the imaging conditions of the first medical image, the method of generating the image conversion model is not limited to this. That is, the method of generating the image conversion model is arbitrary. For example, the first medical image captured under the imaging conditions optimal for the input image IM is input to a model created using a Generative Adversarial Network (GAN) to create a medical image that looks like it was captured under imaging conditions different from the optimal imaging conditions. The image conversion model may be generated by machine learning based on the pseudo-created medical image.

[0030] In this embodiment, since the first modality device 10 and the third modality device 30 are devices of the same manufacturer, an image conversion method for the input image IM can be prepared in advance for each shooting condition of the first medical image captured by the first modality device 10, and the first medical image captured by the first modality device 10 can be converted into the input image IM using the image conversion method prepared in advance. In addition, in the conversion of the second medical image into the input image IM, since the second modality device 20 and the third modality device 30 are modality devices of different manufacturers, it is difficult to prepare in advance an image conversion method for the input image IM for each shooting condition of the second medical image captured by the second modality device 20. Therefore, for the conversion of the second medical image into the input image IM, an image conversion method for the input image IM for each shooting condition of the first medical image captured by the first modality device 10 is used. In the following description, an example will be described in which the image conversion method is an image conversion model.

[0031] The image conversion method information is information on the image conversion method used when converting the first medical image into the input image IM. The image conversion method information includes, for example, image conversion model identification information, which is information for identifying an image conversion model used when converting the first medical image into the input image IM, an image conversion model name, which is information indicating the name of the image conversion model, an adjustment amount, which is information indicating the amount by which the noise and contrast of the first medical image or the second medical image is adjusted when the image conversion model is applied to the first medical image or the second medical image, and used image ratio information regarding the ratio of images used when generating the image conversion method. The used image ratio information includes information such as an image conversion model breakdown, which is information regarding the ratio of types of images such as clinical images and phantom images used when generating the image conversion model, information regarding the accuracy of images for each imaging device that captured the images used when generating the image conversion model, information regarding the ratio of the image range of the images used when generating the image conversion model, and information regarding the ratio of image conditions used when generating the image conversion model. The image conversion method information may include information other than the image conversion model identification information, the image conversion model name, the adjustment amount, and the used image ratio information. In addition, the usage ratio information may include information other than the image model breakdown, information regarding the image accuracy for each photographing device that captured the images used when generating the image transformation model, information regarding the ratio of the image range of the images used when generating the image transformation model, and information regarding the ratio of the image conditions used when generating the image transformation model.

[0032] In this embodiment, the trained model 210 is stored in the memory 51 of the medical image processing device 50. For this reason, in this embodiment, the medical image processing device 50 performs image analysis using the trained model 210. However, the image analysis is not limited to being performed in the medical image processing device 50. That is, the location where the image analysis is performed is arbitrary. For example, the trained model 210 may be stored in a diagnostic support device connected to the medical image processing device 50 via a network NW, and the medical image processing device 50 may transmit the input image IM on which the image conversion process has been performed to the diagnostic support device via the network NW, thereby performing image analysis using the trained model 210 stored in the diagnostic support device.

[0033] FIG. 4 is a diagram showing an example of a first correspondence table in which the imaging conditions of raw data collected in the first modality device 10 are associated with the image conversion conditions in the medical image processing system 1 according to this embodiment. As shown in FIG. 4, in the first correspondence table T1 according to this embodiment, the imaging conditions when the raw data was collected in the first modality device 10 are associated with the image conversion conditions and stored. In the example of the first correspondence table T1 shown in FIG. 4, the medical image processing device 50 associates and stores image conversion conditions RE1 to RE6 with each imaging condition A1 to F1 of the raw data. For example, when the medical image processing device 50 acquires raw data under the imaging condition A1, it acquires the image conversion condition RE1 from the first correspondence table T1 and converts the raw data under the imaging condition A1 into an input image IM using the image conversion condition RE1.

[0034] FIG. 5 is a diagram showing an example of a second correspondence table in which the photographing conditions of the first medical image photographed in the first modality device 10 correspond to image conversion method information in the medical image processing system 1 according to this embodiment. As shown in FIG. 5, for example, in the second correspondence table T2 according to this embodiment, the photographing conditions of the first medical image used when the first medical image was photographed in the first modality device 10, and the image conversion model identification information, the image conversion model name, the adjustment amount, and the image conversion model details as the used image ratio information are associated and stored. In addition, in the second correspondence table T2 according to this embodiment, the photographing conditions of the first medical image and the use ratio information are associated and stored. In the example shown in FIG. 5, the first medical image under the photographing condition A2 is an input image IM. That is, since the first medical image under the photographing condition A2 is an optimal image for the trained model 210, when the medical image processing device 50 acquires the first medical image under the photographing condition A2, the acquired first medical image under the photographing condition A2 can be input to the trained model 210 without conversion. Therefore, the column of the photographing condition A2 of the first medical image in the second correspondence table T2 is not associated with information other than the image conversion model identification information. Note that the second correspondence table T2 may include other information.

[0035] The image conversion model identification information is information for identifying an image conversion model used when converting the first medical image into an input image IM. The imaging conditions of the first medical image and the image conversion model identification information are stored in association with each other in the second correspondence table T2, so that the image conversion model is stored in association with the imaging conditions of the first medical image. As shown in FIG. 5, taking the imaging conditions B2 of the first medical image as an example, the image conversion model identification information is associated with "MO3", taking the imaging conditions C2 of the first medical image as an example, the image conversion model identification information is associated with "MO2", and taking the imaging conditions H2 of the first medical image as an example, the image conversion model identification information is associated with "MO2, MO3". That is, in the example shown in Fig. 4, in the case of the imaging condition B2 of the first medical image, it is indicated that the image conversion model MO3 is applied to the first medical image, in the case of the imaging condition C2 of the first medical image, it is indicated that the image conversion model MO2 is applied to the first medical image, and in the case of the imaging condition H2, it is indicated that the image conversion model MO2 and the image conversion model MO3 are applied to the first medical image. Note that in the example shown in Fig. 5, one or two image conversion models are associated with the imaging condition of one first medical image, but the number of image conversion models for the imaging condition of one first medical image is not limited to one or two. That is, the number of image conversion models for the imaging condition of one first medical image is arbitrary, and three or more image conversion models may be associated with the imaging condition of one first medical image.

[0036] The image conversion model name is information indicating the name of the image conversion model. As shown in Fig. 5, in the case of the first medical image photographing condition B2, the image conversion model name is associated with "noise conversion model A", and in the case of the first medical image photographing condition C2, the image conversion model is associated with "contrast conversion model A".

[0037] The adjustment amount is information indicating the amount by which the noise and contrast of the first medical image are adjusted when the image transformation model is applied to the first medical image. As shown in FIG. 5, taking the imaging condition B2 of the first medical image as an example, the adjustment amount is associated with "1", which indicates that the noise removal is adjusted to increase by 1. Also, taking the imaging condition C2 of the first medical image as an example, the adjustment amount is associated with "1", which indicates that the contrast is adjusted to increase by 1.

[0038] The image conversion model details are information regarding the ratio of the types of images used when generating the image conversion model. The types of images used when generating the image conversion model are clinical images, clinical images (processed), phantoms, etc. A clinical image is an image of a subject, a clinical image (processed) is an image of a subject that has been processed in various ways, and a phantom is an image of a phantom. The greater the ratio of clinical images used, the higher the reliability of the image conversion model, and the greater the ratio of phantoms used, the lower the reliability of the image conversion model. The medical image processing device 50 can present the reliability of the image conversion model used in the image conversion process to the user by presenting this image conversion model details to the user. As shown in Figure 5, taking the first medical image shooting condition B2 as an example, the image conversion model breakdown is associated with "clinical image: 85%, clinical image (processed): 10%, phantom: 5%", and taking the first medical image shooting condition C2 as an example, the image conversion model breakdown is associated with "clinical image: 60%, clinical image (processed): 30%, phantom: 10%".

[0039] The usage ratio information is information on the usage ratio of the image conversion model used when converting to the input image IM. As shown in Fig. 5, taking the imaging condition H2 of the first medical image as an example, it is associated with "2:3", which indicates that the contrast conversion model A and the noise conversion model A are applied at a usage ratio of 2:3. Also, taking the imaging condition B2 of the first medical image as an example, there is only one image conversion model used for the first medical image of the imaging condition B2, so no usage ratio information is associated.

[0040] A case where an image conversion model is applied to a first medical image in this medical image processing device 50 will be described in more detail. FIG. 6 is a diagram showing an example of a case where an image conversion model is applied to a first medical image in the medical image processing device 50 according to this embodiment. As shown in FIG. 6, when the medical image processing device 50 acquires a first medical image FM1 under the shooting condition D2, the medical image processing device 50 applies the high image quality conversion model A, which is the image conversion model MO1 associated with the shooting condition D2 of the first medical image, to the first medical image FM1 under the shooting condition D2 using the second correspondence table T2. This allows the first medical image FM1 under the shooting condition D2 to be converted into an input image IM1. Here, in this embodiment, high image quality means, for example, adjusting both noise and contrast.

[0041] Also, as shown in FIG. 6, when the medical image processing device 50 acquires the first medical image FM2 under the photographing condition C2, the medical image processing device 50 applies the contrast conversion model A, which is the image conversion model MO2 associated with the photographing condition C2 of the first medical image, to the first medical image FM2 under the photographing condition C2 using the second correspondence table T2. This allows the first medical image FM2 under the photographing condition C2 to be converted into the input image IM2. Furthermore, as shown in FIG. 6, when the medical image processing device 50 acquires the first medical image FM3 under the photographing condition B2, the medical image processing device 50 applies the noise conversion model A, which is the image conversion model MO3 associated with the photographing condition B2 of the first medical image FM3, to the first medical image FM3 under the photographing condition B2 using the second correspondence table T2. This allows the first medical image FM3 under the photographing condition B2 to be converted into the input image IM3.

[0042] 5 and 6, the image transformation model according to the present embodiment is an image transformation model that reduces noise, contrast, and improves image quality, but the type of image transformation model is not limited to this. In other words, the type of image transformation model to be created is arbitrary, and for example, image transformation models related to parts, body shapes, contrast conditions, etc. may be stored in the memory 51.

[0043] Returning to FIG. 3, the display 52 outputs various images and information. For example, the display 52 outputs raw data, the first medical image FM, the second medical image, the third medical image, and the like acquired from the first modality device 10, the second modality device 20, the third modality device 30, and the medical data server 40, outputs an input image IM obtained by converting the raw data, the first medical image, the second medical image, and the like, and outputs a GUI (Graphical User Interface) and the like for accepting various operations from a user. In this embodiment, the display 52 is configured by, for example, a liquid crystal display, a CRT (Cathode Ray Tube) display, and the like. This display 52 corresponds to the display unit in this embodiment.

[0044] The input interface 53 is a circuit that accepts various inputs from a user, and in this embodiment, the input interface 53 accepts, for example, input operations from the user regarding the selection of medical data MD to be converted, adjustment of the usage ratio, etc. In this embodiment, the input interface 53 is realized by, for example, a mouse, a keyboard, a trackball, a manual switch, a foot switch, a button, a joystick, etc.

[0045] The communication interface 54 implements various information communication protocols according to the form of the network NW. The communication interface 54 realizes communication with other devices via the network NW according to these various protocols. In particular, in this embodiment, the medical image processing device 50 is connected to the network NW via the communication interface 54, and communication with the first modality device 10, the second modality device 20, the third modality device 30, and the medical data server 40 is realized.

[0046] The processing circuit 55 is a calculation circuit that performs various calculations, and is configured with a processor such as a CPU, a GPU, etc. The processing circuit 55 according to the present embodiment converts medical data MD into an input image IM, and outputs an analysis result of an image output by the trained model 210, for example.

[0047] For this reason, the processing circuit 55 according to this embodiment has an acquisition function 551, a conversion function 552, an output function 553, a first identification function 554, a calculation function 555, a filtering function 556, a selection function 557, an adjustment function 558, and a registration function 559. The acquisition function 551 corresponds to the acquisition unit in this embodiment, the conversion function 552 corresponds to the conversion unit in this embodiment, the output function 553 corresponds to the output unit in this embodiment, the first identification function 554 corresponds to the first identification unit in this embodiment, the calculation function 555 corresponds to the calculation unit in this embodiment, the filtering function 556 corresponds to the filtering unit in this embodiment, the selection function 557 corresponds to the selection unit in this embodiment, the adjustment function 558 corresponds to the adjustment unit in this embodiment, and the registration function 559 corresponds to the registration unit in this embodiment.

[0048] In the embodiment shown in Fig. 3, the processing functions performed by the acquisition function 551, the conversion function 552, the output function 553, the first identification function 554, the calculation function 555, the filtering function 556, the selection function 557, the adjustment function 558, and the registration function 559 are stored in the memory 51 in the form of a program executable by a computer. The processing circuit 55 is a processor that realizes a function corresponding to each program by reading the program from the memory 51 and executing it. In other words, the processing circuit 55 in a state in which each program has been read has each function shown in the processing circuit 55 of Fig. 3. In FIG. 3, it has been described that the acquisition function 551, the conversion function 552, the output function 553, the first identification function 554, the calculation function 555, the filtering function 556, the selection function 557, the adjustment function 558, and the registration function 559 are realized by a single processing circuit 55. However, it is also possible to configure the processing circuit 55 by combining a plurality of independent processors, and realize these functions by each processor executing a program.

[0049] The acquisition function 551 acquires medical data MD. The conversion function 552 converts the medical data MD acquired by the acquisition function 551 into an input image IM that is suitable for the input of the trained model 210 that outputs the analysis result of the image. Specifically, when the acquisition function 551 acquires raw data as the medical data MD, the conversion function 552 acquires image conversion conditions corresponding to the shooting conditions of the raw data, and performs image conversion processing using the acquired image conversion conditions on the raw data, thereby converting the raw data into an input image IM. In addition, when the acquisition function 551 acquires a first medical image or a second medical image as the medical data MD, the conversion function 552 converts the first medical image or the second medical image into an input image IM by performing image conversion processing on the first medical image or the second medical image using an image conversion model in the image conversion method stored in the memory 51.

[0050] The output function 553 outputs the input image IM converted by the conversion function 552. The first identification function 554 uses the specific model to identify an image conversion method used in the image conversion processing of the first medical image. In this embodiment, the first identification function 554 uses the specific model to identify an image conversion model used in the image conversion processing of the first medical image. Here, the specific model is a model that identifies an image conversion model used in the image conversion processing of the medical image based on the medical image. Specifically, the specific model is a model that identifies an image conversion model used in image conversion for an input medical image by classification or the like.

[0051] The image transformation model identified in this specific model is not limited to one, and multiple image transformation models may be identified based on the shooting conditions, etc. In addition, when multiple image transformation models are identified using the specific model, the usage ratio information of the identified multiple image transformation models is also identified.

[0052] The calculation function 555 calculates the imaging conditions of the second medical image in the first modality device 10 based on the second medical image captured in the second modality device 20 different from the first modality device 10. Specifically, when converting the second medical image into the input image IM, the imaging conditions of the first medical image in the first modality correspond to the image conversion model, so that the imaging conditions of the second medical image in the second modality device 20 cannot specify the image conversion model used when converting to the input image IM. Therefore, the calculation function 555 calculates the imaging conditions of the second medical image in the first modality device 10. By the calculation function 555 calculating the imaging conditions of the second medical image in the first modality device 10 in this way, the conversion function 552 can use the second correspondence table T2 to specify the image conversion model used when converting to the input image IM and convert the second medical image into the input image IM.

[0053] Here, the calculation of the shooting conditions of the second medical image in the first modality device 10 is performed, for example, by using a second medical image whose shooting conditions in the first modality device 10 are unknown, and adjusting the shooting conditions of the first medical image whose shooting conditions in the first modality device 10 are known.

[0054] 7 and 8 are explanatory diagrams for explaining an example of a calculation method of the imaging conditions of the first modality device 10 of the second medical image in the medical image processing device 50 according to this embodiment. In the example shown in FIG. 7, the calculation function 555 acquires the first medical image FM whose imaging conditions in the first modality device 10 are known and the second medical image SM whose imaging conditions in the first modality device 10 are to be calculated, and outputs the first medical image FM, the second medical image SM, and an operation area OP1 for receiving an input operation for specifying the imaging conditions of the first medical image FM to the display 52. ​​Next, the calculation function 555 adjusts the imaging conditions of the first medical image FM according to the input operation on the slide bar SB1 while outputting to the display 52 an estimated image estimated to be captured by the first modality device 10 based on the imaging conditions of the first medical image FM after adjustment according to the input operation on the slide bar SB1, and the second medical image SM captured in the second modality device 20. In addition, the estimated image is output to the display 52 using, for example, a first model that has learned the first medical image FM and the shooting conditions of the first medical image FM, and an estimated image corresponding to the shooting conditions in the first modality device 10 after adjustment in accordance with the input operation.

[0055] Then, as shown in Figure 8, the calculation function 555 calculates the shooting conditions of the estimated image FM' output to the operation area OP1 when the estimated image FM' becomes equivalent to the second medical image SM as the shooting conditions of the second medical image SM in the first modality device 10.

[0056] In addition, the calculation of the shooting conditions in the first modality device 10 of the second medical image may be performed automatically, for example, by comparing the similarity between the first medical image FM, whose shooting conditions in the first modality device 10 are known, and the second medical image SM, whose shooting conditions in the first modality device 10 are to be calculated.

[0057] FIG. 9 is an explanatory diagram for explaining another example of a calculation method of the imaging conditions of the second medical image SM in the first modality device 10 in the medical image processing device 50 according to this embodiment. As shown in FIG. 9, the calculation function 555 acquires a plurality of first medical images FM and a second medical image SM for which the imaging conditions in the first modality device 10 are to be calculated. Next, the calculation function 555 compares the plurality of first medical images FM with the second medical image SM for which the imaging conditions in the first modality device 10 are to be calculated, and determines the similarity between each of the plurality of first medical images FM and the second medical image SM for which the imaging conditions in the first modality device 10 are to be calculated. Then, the calculation function 555 may calculate the imaging conditions of the second medical image SM in the first modality device 10 by estimating the imaging conditions of the first medical image FM of the plurality of first medical images FM having a similarity equal to or greater than a threshold value as the imaging conditions of the second medical image SM in the first modality device 10.

[0058] In addition, the calculation of the shooting conditions in the first modality device 10 for the second medical image SM may be performed automatically using a second model that has learned the images obtained when the values ​​of each shooting condition are changed for each of multiple types of shooting conditions of the first modality device 10.

[0059] 10 is an explanatory diagram for explaining another example of a calculation method of imaging conditions in the first modality device 10 of the second medical image SM in the medical image processing device 50 according to this embodiment. As shown in FIG. 10, this second model is a model that learns an image estimated when the value of the tube voltage, which is one of the imaging conditions, is changed, and an image estimated when the value of the reconstruction function, which is another type of imaging condition, is changed. In the example shown in FIG. 10, the shading of the estimated image is changed when the tube voltage is changed. Also, when the reconstruction function is changed, the color of the estimated image is changed.

[0060] Using this second model, the calculation function 555 selects one type of imaging condition from the second model among the multiple types of imaging conditions. Next, while changing the value of the selected one type of imaging condition, the calculation function 555 compares an image corresponding to the value of the one type of imaging condition acquired from the second model with the second medical image SM for which the imaging condition in the first modality device 10 is to be calculated, and determines the similarity. Then, the calculation function 555 repeats the process of determining one type of imaging condition having a value when the similarity becomes equal to or greater than a threshold value until all values ​​of the multiple types of imaging conditions are determined. In this way, the calculation function 555 may estimate the imaging condition in the first modality device 10 of the second medical image SM for each of the multiple types of imaging conditions, and calculate the imaging condition in the first modality device 10 of the second medical image SM.

[0061] The method of calculating the imaging conditions of the second medical image SM in the first modality apparatus 10 is not limited to the above example. That is, the method of calculating the imaging conditions of the second medical image SM in the first modality apparatus 10 is arbitrary.

[0062] The filtering function 556 filters the imaging conditions of the first medical image FM stored in the memory 51 based on the imaging conditions of the second medical image SM in the first modality device 10 calculated by the calculation function 555. The selection function 557 selects an image conversion model to be used for image conversion processing of the second medical image SM from the image conversion models associated with the imaging conditions of the first medical image FM filtered by the filtering function 556 based on the second medical image SM and the input image IM.

[0063] The adjustment function 558 receives an input operation for adjusting the usage ratio from the user, and adjusts the usage ratio based on the received input operation. The registration function 559 registers the image conversion method information and the shooting conditions in the memory 51 in association with each other.

[0064] 11 and 12 are flow charts for explaining the contents of the conversion process executed by the medical image processing device 50 according to this embodiment. In the conversion process, medical data MD is acquired, the medical data MD is converted into an input image IM, and the converted input image IM is output to the trained model 210. For example, this conversion process is executed when a user instructs it to be started.

[0065] 11, first, the medical image processing device 50 acquires medical data MD (step S11). The process of acquiring the medical data MD is realized by an acquisition function 551 in the processing circuit 55. Specifically, the medical image processing device 50 acquires, as medical data MD, any one of raw data collected in the first modality device 10, a first medical image FM captured in the first modality device 10, and a second medical image SM captured in the second modality device 20 from the first modality device 10, the second modality device 20, or the medical data server 40.

[0066] 11, the medical image processing device 50 acquires the additional information of the medical data MD (step S13). The process of acquiring the additional information of the medical data MD is realized by the acquisition function 551 in the processing circuit 55. Specifically, the medical image processing device 50 acquires the additional information of the medical data MD acquired in step S11 together with the medical data MD from the first modality device 10, the second modality device 20, or the medical data server 40.

[0067] Here, the additional information is information that is added to the medical data MD and is generated by the modality that imaged the subject. The additional information includes, for example, information on the imaging conditions of the medical data MD, information on the manufacturer of the modality that generated the medical data MD, information on the type of data of the medical data MD such as raw data or medical images, etc.

[0068] The incidental information may be acquired together with the medical data MD in step S11. If the incidental information is acquired together with the medical data MD in step S11, the process of acquiring the incidental information in step S13 may be omitted.

[0069] 11, the medical image processing device 50 identifies the medical data MD (step S15). This process of identifying the medical data MD is realized by the acquisition function 551 in the processing circuit 55. Specifically, the medical image processing device 50 identifies whether the medical data MD acquired in step S11 is raw data, a first medical image FM, or a second medical image SM, based on the accompanying information acquired in step S13.

[0070] 11, the medical image processing device 50 determines whether or not the medical data MD has been captured by the first modality device 10 (step S17). The process of determining whether or not the medical data MD has been captured by the first modality device 10 is realized by the conversion function 552 in the processing circuit 55. Specifically, the medical image processing device 50 determines whether or not the medical data MD has been captured by the first modality device 10 based on the result of identifying the medical data MD in step S15.

[0071] Then, in step S17, if the medical data MD is captured by the first modality device 10 (step S17: YES), the medical image processing device 50 determines whether or not the medical data MD is raw data (step S19). The process of determining whether or not the medical data MD is raw data is realized by the conversion function 552 in the processing circuit 55. Specifically, the medical image processing device 50 determines whether or not the medical data MD is raw data based on the result of identifying the medical data MD in step S15.

[0072] Then, in step S19, if the medical data MD is raw data (step S19: YES), that is, if the acquisition function 551 acquires raw data as the medical data MD, the medical image processing device 50 acquires image conversion conditions corresponding to the imaging conditions of the raw data (step S21). The process of acquiring the image conversion conditions corresponding to the imaging conditions of the raw data is realized by the conversion function 552 in the processing circuit 55. Specifically, the medical image processing device 50 acquires the image conversion conditions corresponding to the imaging conditions of the raw data using the first correspondence table T1. More specifically, the conversion function 552 identifies the image conversion conditions corresponding to the imaging conditions of the raw data using the first correspondence table T1, and acquires the identified image conversion conditions from the memory 51.

[0073] 11, the medical image processing apparatus 50 converts the raw data into an input image IM (step S23). This process of converting the raw data into the input image IM is realized by the conversion function 552 in the processing circuitry 55. Specifically, the medical image processing apparatus 50 converts the raw data into the input image IM by executing image conversion processing on the raw data using the image conversion conditions acquired in step S21. More specifically, when the reconstruction conditions are acquired as the image conversion conditions in step S21, the medical image processing apparatus 50 converts the raw data into the input image IM by executing reconstruction processing on the raw data based on the acquired reconstruction conditions.

[0074] 11, the medical image processing apparatus 50 outputs the input image IM to the trained model 210 (step S25). The process of outputting the input image IM to the trained model 210 is realized by the output function 553 in the processing circuitry 55. Specifically, the medical image processing apparatus 50 outputs the input image IM converted by the conversion function 552 in step S23 to the trained model 210 stored in the memory 51.

[0075] 11, the medical image processing apparatus 50 acquires the analysis result AR (step S27). The process of acquiring the analysis result AR is realized by the acquisition function 551 in the processing circuitry 55. Specifically, the medical image processing apparatus 50 acquires the analysis result AR from the trained model 210.

[0076] 11, the medical image-processing device 50 outputs the analysis result AR (step S29). The process of outputting the analysis result AR is realized by the output function 553 in the processing circuitry 55. Specifically, the medical image-processing device 50 outputs the analysis result AR acquired in step S27 to the display 52, and ends the conversion process.

[0077] On the other hand, in step S19, when the medical data MD is not raw data (step S19: NO), that is, when the acquisition function 551 acquires the first medical image FM as the medical data MD, the medical image processing device 50 judges whether or not the imaging conditions of the first medical image FM are stored (step S31). The process of judging whether or not the imaging conditions of the first medical image FM are stored is realized by the conversion function 552 in the processing circuit 55. Specifically, the medical image processing device 50 judges whether or not the imaging conditions of the first medical image FM acquired by the acquisition function 551 are stored in the memory 51. More specifically, the conversion function 552 judges whether or not the imaging conditions of the first medical image FM are stored in the second correspondence table T2 of the memory 51 by using the second correspondence table T2 stored in the memory 51.

[0078] Then, in step S31, when it is determined that the imaging conditions of the first medical image FM are stored (step S31: YES), that is, when the imaging conditions of the first medical image FM are stored in the memory 51, the medical image processing device 50 acquires an image conversion model corresponding to the imaging conditions of the first medical image FM (step S33). The process of acquiring the image conversion model corresponding to the imaging conditions of the first medical image FM is realized by the conversion function 552 in the processing circuit 55. Specifically, the medical image processing device 50 acquires an image conversion model corresponding to the imaging conditions of the first medical image FM using the second correspondence table T2. More specifically, the conversion function 552 identifies an image conversion model corresponding to the imaging conditions of the first medical image FM using the second correspondence table T2, and acquires the identified image conversion model from the memory 51. Note that in this step S33, when a plurality of image conversion models are identified using the identified model, the usage ratio information of the identified plurality of image conversion models is also identified.

[0079] 11, the medical image processing apparatus 50 converts the first medical image FM into an input image IM using the acquired image conversion model (step S35). This process of converting the first medical image FM into the input image IM is realized by the conversion function 552 in the processing circuitry 55. Specifically, the medical image processing apparatus 50 converts the first medical image FM into the input image IM by executing image conversion processing on the first medical image FM using the image conversion model acquired in step S33.

[0080] On the other hand, when it is determined in step S31 that the imaging conditions of the first medical image FM are not stored (step S31: NO), that is, when the imaging conditions of the first medical image FM are not stored in the memory 51, the medical image processing device 50 identifies an image conversion model corresponding to the first medical image FM (step S37). The process of identifying the image conversion model corresponding to the first medical image FM is realized by the first identification function 554 in the processing circuitry 55. Specifically, the medical image processing device 50 inputs the first medical image FM to the identification model, thereby identifying the image conversion model used in the image conversion process of the first medical image FM.

[0081] 11, the medical image processing apparatus 50 converts the first medical image FM into an input image IM using the specified image conversion model (step S39). This process of converting the first medical image FM into the input image IM is realized by the conversion function 552 in the processing circuitry 55. Specifically, the medical image processing apparatus 50 converts the first medical image FM into the input image IM by executing image conversion processing on the first medical image FM using the image conversion model specified in step S37.

[0082] On the other hand, in step S17, if the medical data MD is not captured by the first modality device 10 (step S17: NO), that is, if the acquisition function 551 acquires the second medical image SM captured by the second modality device 20 different from the first modality device 10 as the medical data MD, the medical image processing device 50 judges whether or not the imaging conditions of the second medical image SM in the second modality device 20 are stored (step S41). The process of judging whether or not the imaging conditions of the second medical image SM in the second modality device 20 are stored is realized by the conversion function 552 in the processing circuit 55. Specifically, the medical image processing device 50 judges whether or not the imaging conditions of the second medical image SM in the second modality device 20 acquired by the acquisition function 551 are stored in the memory 51. More specifically, the conversion function 552 uses the second correspondence table T2 stored in the memory 51 to determine whether the shooting conditions of the second medical image SM in the second modality device 20 are stored in the second correspondence table T2 of the memory 51.

[0083] Then, in step S41, if the imaging conditions of the second medical image SM in the second modality device 20 are stored (step S41: YES), the medical image processing device 50 acquires an image transformation model corresponding to the imaging conditions of the second medical image SM in the second modality device 20 (step S43). The process of acquiring the image transformation model corresponding to the imaging conditions of the second medical image SM in the second modality device 20 is realized by the conversion function 552 in the processing circuit 55. Specifically, the medical image processing device 50 acquires an image transformation model corresponding to the imaging conditions of the second medical image SM in the second modality device 20 using the second correspondence table T2. More specifically, the conversion function 552 identifies an image transformation model corresponding to the imaging conditions of the second medical image SM in the second modality device 20 using the second correspondence table T2, and acquires the identified image transformation model from the memory 51.

[0084] 11, the medical image processing apparatus 50 converts the second medical image SM into an input image IM using the acquired image conversion model (step S45). This process of converting the second medical image SM into the input image IM is realized by the conversion function 552 in the processing circuitry 55. Specifically, the medical image processing apparatus 50 converts the second medical image SM into the input image IM by executing image conversion processing on the second medical image SM using the image conversion model acquired in step S43.

[0085] On the other hand, in step S41, if the imaging conditions of the second medical image SM in the second modality device 20 are not stored (step S41: NO), the medical image processing device 50 calculates the imaging conditions of the first modality device 10 of the second medical image SM (step S47). The process of calculating the imaging conditions of the first modality device 10 of the second medical image SM is realized by the calculation function 555 in the processing circuitry 55. Specifically, the medical image processing device 50 calculates the imaging conditions of the first modality device 10 of the second medical image SM based on the second medical image SM.

[0086] Next, as shown in FIG. 11, the medical image processing device 50 judges whether the imaging conditions of the second medical image SM in the first modality device 10 are stored (step S49). The process of judging whether the imaging conditions of the second medical image SM in the first modality device 10 are stored is realized by the conversion function 552 in the processing circuit 55. Specifically, the medical image processing device 50 judges whether the imaging conditions of the second medical image SM in the first modality device 10 calculated by the calculation function 555 are stored in the memory 51. More specifically, the conversion function 552 judges whether the imaging conditions of the second medical image SM in the first modality device 10 are stored in the second correspondence table T2 of the memory 51 by using the second correspondence table T2 stored in the memory 51.

[0087] Then, in step S49, if the imaging conditions of the second medical image SM in the first modality device 10 are stored (step S49: YES), that is, if the imaging conditions of the second medical image SM in the first modality device 10 are stored in the memory 51, the medical image processing device 50 acquires an image transformation model corresponding to the imaging conditions of the second medical image SM in the first modality device 10 (step S51). The process of acquiring the image transformation model corresponding to the imaging conditions of the second medical image SM in the first modality device 10 is realized by the conversion function 552 in the processing circuit 55. Specifically, the medical image processing device 50 acquires an image transformation model corresponding to the imaging conditions of the second medical image SM in the first modality device 10 using the second correspondence table T2. More specifically, the conversion function 552 identifies an image transformation model corresponding to the imaging conditions of the second medical image SM in the first modality device 10 using the second correspondence table T2, and acquires the identified image transformation model from the memory 51.

[0088] 11, the medical image processing apparatus 50 converts the second medical image SM into an input image IM using the acquired image conversion model (step S53). This process of converting the second medical image SM into the input image IM is realized by the conversion function 552 in the processing circuitry 55. Specifically, the medical image processing apparatus 50 converts the second medical image SM into the input image IM by executing image conversion processing on the second medical image SM using the image conversion model acquired in step S51.

[0089] On the other hand, in step S49, if the imaging conditions of the first modality device 10 of the second medical image SM are not stored (step S49: NO), that is, if the imaging conditions of the first modality device 10 of the second medical image SM are not stored in the memory 51, the medical image processing device 50 filters the imaging conditions of the first medical image FM (step S55). The process of filtering the imaging conditions of the first medical image FM is realized by the filtering function 556 in the processing circuitry 55. Specifically, the medical image processing device 50 filters the imaging conditions of the first medical image FM based on the imaging conditions of the first modality device 10 of the second medical image SM calculated in step S47. More specifically, the filtering function 556 filters the shooting conditions of the first medical image FM that include the same shooting area and shooting protocol name as the shooting area and shooting protocol name included in the shooting conditions in the first modality device 10 of the second medical image SM from the shooting conditions of the first medical image FM associated with the second correspondence table T2.

[0090] Next, as shown in FIG. 11, the medical image processing device 50 selects an image conversion model (step S57). This image conversion model selection process is realized by the selection function 557 in the processing circuit 55. Specifically, the medical image processing device 50 selects an image conversion model to be used for image conversion processing of the second medical image SM from the image conversion models associated with the shooting conditions of the first medical image FM filtered by the filtering unit, based on the second medical image SM and the input image IM. More specifically, the selection function 557 acquires the input image IM for the second medical image SM, evaluates the difference in image quality such as contrast and noise between the second medical image SM and the input image IM, compares the evaluation result with the image conversion method information associated with the second correspondence table T2, and selects the image conversion model.

[0091] For example, in step S57, when the selection function 557 obtains an evaluation result that the noise is insufficient by 0.5 and the contrast is insufficient by 1 in order to convert the second medical image SM into the input image IM by evaluating the difference in image quality between the second medical image SM and the input image IM, the selection function 557 refers to the image conversion method information associated with the second correspondence table T2. Then, since there is no image conversion model that increases the noise by 0.5, but an image conversion model that increases the contrast by 1 is associated with the second correspondence table T2, the selection function 557 selects the contrast conversion model A, which is the image conversion model MO2, based on the evaluation result. Note that in this step S57, the image conversion model selected is not limited to one, and multiple image conversion models may be selected. The imaging conditions in the first modality device 10 of this second medical image SM and the imaging conditions of the first medical image FM selected by the selection function 557 may be called similar.

[0092] 11, the medical image processing apparatus 50 converts the second medical image SM into an input image IM using the selected image conversion model (step S59). This process of converting the second medical image SM into the input image IM is realized by the conversion function 552 in the processing circuitry 55. Specifically, the medical image processing apparatus 50 converts the second medical image SM into the input image IM by executing image conversion processing on the second medical image SM using the image conversion model selected in step S57.

[0093] 12, the medical image-processing device 50 outputs the input image IM to the display 52 (step S61). This process of outputting the input image IM to the display 52 is realized by the output function 553 in the processing circuitry 55. Specifically, the medical image-processing device 50 outputs the input image IM converted in step S35, step S39, step S45, step S53, or step S59 to the display 52.

[0094] 12, the medical image processing apparatus 50 acquires image conversion method information and usage ratio information (step S63). The process of acquiring the image conversion method information is realized by the acquisition function 551 in the processing circuit 55. Specifically, the medical image processing apparatus 50 acquires image conversion method information corresponding to the image conversion method used when converting the image in step S35, step S39, step S45, step S53, or step S59, and usage ratio information of the image conversion model used in step S35, step S39, step S45, step S53, or step S59. Note that, in step S63, when only one image conversion model is used when converting into the input image IM, only the image conversion method information is acquired.

[0095] 12, the medical image processing apparatus 50 determines whether or not multiple image transformation models have been used (step S64). The process of determining whether or not multiple image transformation models have been used is realized by the adjustment function 558 in the processing circuitry 55. Specifically, the medical image processing apparatus 50 determines whether or not multiple image transformation models have been used when transforming the input image IM, based on the image transformation method information acquired in step S63.

[0096] Then, in step S64, if it is determined that a plurality of image transformation models are used (step S64: YES), the medical image processing apparatus 50 outputs an operation area including a slider bar (step S65). The process of outputting the operation area including the slider bar is realized by the adjustment function 558 in the processing circuitry 55. Specifically, the medical image processing apparatus 50 outputs to the display 52 an operation area including a slider bar that accepts an input operation for adjusting the usage ratio of the image transformation model based on the image transformation method information and the usage ratio information acquired in step S63.

[0097] The output mode of the input image IM and the operation area will be described with reference to Figs. 13 to 15. Fig. 13 is a diagram showing an example of the output mode of the input image IM and the operation area output to the display 52 of the medical image processing device 50 according to this embodiment. Figs. 14 and 15 are diagrams showing an example of the selection operation of the image conversion model in the operation area in the medical image processing device 50 according to this embodiment. As shown in Fig. 13, the medical image processing device 50 outputs the input image IM to the display 52 and outputs the operation area OP2 below the input image IM. In addition, the medical image processing device 50 outputs a pull-down PD for selecting the image conversion model and a slider bar SB2 capable of adjusting the usage ratio of the image conversion model to the operation area OP2. In the example shown in Fig. 13, the contrast conversion model A is selected in the pull-down PD, so the usage ratio of the contrast conversion model A can be increased or decreased by operating the slider bar SB2. Specifically, for example, when the user operates the slider bar SB2 in the d1 direction, the usage ratio of contrast conversion model A increases, and when the user operates the slider bar SB2 in the direction opposite to the d1 direction, the usage ratio of contrast conversion model A decreases.

[0098] Also, as shown in FIG. 14, when the user clicks on the pull-down PD in the operation area OP2 via the input interface 53, a pull-down menu presenting selectable image conversion models is output in the operation area OP2. In the example shown in FIG. 14, since a plurality of image conversion models are used when converting into an input image IM, contrast conversion model A, contrast conversion model B, contrast conversion model C, and contrast conversion model D are output in a selectable manner in the pull-down menu. In the example shown in FIG. 14, when contrast conversion model B is selected, the usage ratio of contrast conversion model B can be increased or decreased by operating slider bar SB2. In addition, when this pull-down menu is output, the user can specify any one of the image conversion models using the mouse pointer MP, and the details of the specified image conversion model are output. In the example shown in FIG. 15, since contrast conversion model A is specified, "clinical image: 85%, clinical image (processed): 10%, phantom: 5%" is output as the image conversion model details R1 of contrast conversion model A. In addition, in the example shown in Figure 15, contrast conversion model B is specified, so the image conversion model breakdown R2 of contrast conversion model A is output as "clinical image: 35%, clinical image (processed): 30%, phantom: 25%."

[0099] 12, the medical image processing apparatus 50 determines whether the usage ratio has been adjusted (step S67). The process of determining whether the usage ratio has been adjusted is realized by the adjustment function 558 in the processing circuitry 55. Specifically, the medical image processing apparatus 50 receives an input operation for adjusting the usage ratio of the image transformation model from the user via the input interface 53, and determines whether the usage ratio of the image transformation model has been adjusted. More specifically, the adjustment function 558 receives an input operation on the slider bar SB2 as the input operation for adjusting the usage ratio, and determines whether the usage ratio has been adjusted.

[0100] Then, in step S67, if the usage ratio is adjusted (step S67: YES), the medical image processing apparatus 50 converts into an input image IM (step S69). This process of converting into an input image IM is realized by the conversion function 552 in the processing circuitry 55. Specifically, the medical image processing apparatus 50 converts the first medical image FM or the second medical image SM into an input image IM based on the usage ratio adjusted in step S67.

[0101] 12, the medical image-processing device 50 outputs the input image IM (step S71). The process of outputting the input image IM is realized by the output function 553 in the processing circuitry 55. Specifically, the medical image-processing device 50 outputs the input image IM based on the adjusted usage ratio converted in step S69 to the display 52.

[0102] On the other hand, in step S67, if the usage ratio is not adjusted (step S67: NO) or after step S71, the medical image-processing device 50 determines whether the adjustment of the usage ratio has been completed (step S73). The process of determining whether the adjustment of the usage ratio has been completed is realized by the adjustment function 558 in the processing circuit 55. Specifically, the medical image-processing device 50 determines whether the adjustment of the usage ratio has been completed by determining whether a button (not shown) for ending the adjustment of the usage ratio has been pressed via the input interface 53. Note that the end of the adjustment of the usage ratio is not limited to when a button has been pressed, and the end of the adjustment of the usage ratio may be determined by inputting a voice via the input interface 53 for ending the change of the image conversion method information.

[0103] Then, in step S73, if the adjustment of the usage ratio is not completed (step S73: NO), the medical image-processing apparatus 50 returns to step S67 and repeats the process from step S67 until the change of the image conversion technique information is completed.

[0104] On the other hand, if it is determined in step S64 that a plurality of image transformation models are used (step S64: NO), or if the adjustment of the usage ratio is completed in step S73 (step S73: YES), the medical image processing apparatus 50 outputs the input image IM to the trained model 210 (step S75). The process of outputting the input image IM to the trained model 210 is realized by the output function 553 in the processing circuitry 55. Specifically, the medical image processing apparatus 50 outputs the input image IM to the trained model 210.

[0105] 12, the medical image-processing device 50 acquires the analysis result AR (step S77). The process of acquiring the analysis result AR is realized by the acquisition function 551 in the processing circuitry 55. Specifically, the medical image-processing device 50 acquires the analysis result AR for the input image IM output by the trained model 210.

[0106] 12, the medical image processing apparatus 50 outputs the analysis result AR, the image conversion method information, and the usage ratio information (step S79). The process of outputting the analysis result AR, the image conversion method information, and the usage ratio information is realized by the output function 553 in the processing circuitry 55. Specifically, in step S75, the medical image processing apparatus 50 outputs the image conversion method information used in converting into the input image IM and the usage ratio information together with the analysis result AR output from the trained model 210 to the display 52. ​​Note that in step S79, when only one image conversion model is used in converting into the input image IM, only the image conversion method information is output to the display 52 together with the analysis result AR.

[0107] FIG. 16 is a diagram showing an example of the analysis result AR, image conversion method information, and usage ratio information outputted to the display 52 of the medical image processing device 50 according to this embodiment. As shown in FIG. 16, the analysis result AR is outputted to the display 52, a report display area RD in which information related to the report is displayed is outputted below the analysis result AR, and an information display area ID is outputted to the right of the analysis result AR. The image conversion model used in converting the input image IM and the usage ratio information are outputted in this information display area ID. In the example shown in FIG. 16, the contrast conversion model indicated as contrast Lv, the noise conversion model indicated as noise adjustment correction, the high image quality conversion model indicated as high image quality, and the other three conversion models are outputted as the image conversion method information related to the image conversion model used in converting the input image IM, and the ratio of the contrast conversion model, the noise conversion model, the high image quality conversion model, and the other three conversion models, "3:2:1:0:5:1", is outputted as the usage ratio information. In this way, by outputting the image transformation method information and the usage ratio information together with the analysis result AR to the display 52, the user can know what image transformation model was used.

[0108] By executing step S79, the conversion process according to this embodiment is completed.

[0109] 17 is a flow chart for explaining the contents of the update process executed by the medical image processing device 50 of the medical image processing system 1 according to this embodiment. In this update process, image conversion method information and imaging conditions are acquired, and when the imaging conditions are not registered or the image conversion method information is updated, the imaging conditions are newly registered or the image conversion method information is updated. For example, this update process is executed when the input image IM is output to the trained model 210.

[0110] 17, first, the medical image processing device 50 determines whether or not the input image IM has been output to the trained model 210 (step S81). The process of determining whether or not the input image IM has been output to the trained model 210 is realized by the registration function 559 in the processing circuitry 55. Then, if the input image IM has not been output to the trained model 210 (step S81: NO), the process of step S81 is repeated and the process waits until the input image IM is output to the trained model 210.

[0111] On the other hand, in step S81, when the input image IM is output to the trained model 210 (step S81: YES), the medical image processing device 50 acquires image conversion method information and imaging conditions of the input image IM (step S83). The process of acquiring the image conversion method information and imaging conditions of the input image IM is realized by the registration function 559 in the processing circuitry 55. Specifically, the medical image processing device 50 acquires the imaging conditions of the first medical image FM or the second medical image SM before being converted into the input image IM and the image conversion method information used when converting into the input image IM.

[0112] 17, the medical image processing apparatus 50 judges whether or not the imaging conditions are registered (step S85). The process of judging whether or not the imaging conditions are registered is realized by the registration function 559 in the processing circuitry 55. Specifically, the medical image processing apparatus 50 judges whether or not the imaging conditions of the first medical image FM or the second medical image SM before being converted into the input image IM are registered in the second correspondence table T2.

[0113] Then, in step S85, if the imaging conditions are not registered (step S85: NO), the medical image-processing device 50 registers the image conversion method information and the imaging conditions in association with each other (step S87). This process of registering the image conversion method information and the imaging conditions in association with each other is realized by the registration function 559 in the processing circuitry 55. Specifically, the image conversion method information and the imaging conditions acquired in step S83 are registered in the memory 51 by registering them in the second correspondence table T2 in association with each other.

[0114] On the other hand, in step S85, if the imaging conditions are registered (step S85: YES), the medical image-processing device 50 determines whether the image conversion method information is different (step S89). The process of determining whether the image conversion method information is different is realized by the registration function 559 in the processing circuit 55. Specifically, the medical image-processing device 50 determines whether the image conversion method information associated with the imaging conditions in the second correspondence table T2 is different from the image conversion method information acquired in step S83.

[0115] Then, in step S89, if the image conversion method information is different (step S89: YES), the medical image-processing device 50 updates the image conversion method information corresponding to the imaging condition (step S91). This process of updating the image conversion method information corresponding to the imaging condition is realized by the registration function 559 in the processing circuitry 55. Specifically, the medical image-processing device 50 updates the image conversion method information associated with the imaging condition in the second correspondence table T2 to the image conversion method information acquired in step S83.

[0116] After step S87, if the image conversion method information is not different in step S89 (step S89: NO), and after step S91, the update process according to this embodiment is terminated. In this update process, if a plurality of image conversion models are used when converting the input image IM, the medical image processing device 50 may acquire usage ratio information together with the image conversion method information and the imaging conditions, and when the imaging conditions are not registered or when the image conversion method information and / or the usage ratio information are updated, may register the image conversion method information and the usage ratio information in association with the imaging conditions, or may update the image conversion method information and / or the usage ratio information.

[0117] As described above, according to the medical image processing device 50 of the medical image processing system 1 of this embodiment, when the medical image processing device 50 acquires raw data as medical data MD, it acquires image conversion conditions corresponding to the shooting conditions of the acquired raw data, and converts the raw data into an input image IM by performing reconstruction processing on the raw data using the acquired image conversion conditions.Therefore, images captured by a different modality device can be applied to the trained model 210 without performing additional learning on the trained model 210.

[0118] Furthermore, according to the medical image processing device 50 of the medical image processing system 1 of this embodiment, when the medical image processing device 50 acquires a first medical image FM as medical data MD, it acquires or identifies an image conversion processing model corresponding to the shooting conditions of the acquired first medical image FM, and converts the first medical image FM into an input image IM using the acquired or identified image conversion model.Therefore, as in the case of raw data, an image captured with a different modality device can be applied to the trained model 210 without performing additional learning on the trained model 210.

[0119] Furthermore, according to the medical image processing device 50 of the medical image processing system 1 of this embodiment, when the medical image processing device 50 acquires a second medical image SM as medical data MD, it acquires an image conversion model corresponding to the shooting conditions of the second medical image SM in the second modality device 20, or calculates the shooting conditions of the second medical image SM in the first modality device 10 from the second medical image SM and acquires an image conversion model corresponding to the shooting conditions of the calculated second medical image SM in the first modality device 10, or selects an image conversion model to be used in the image conversion processing of the second medical image SM, and converts the second medical image SM into an input image IM using the acquired or selected image conversion model.Therefore, as in the case of raw data and the first medical image FM, images captured with a different modality device can be applied to the trained model 210 without performing additional learning on the trained model 210.

[0120] In the above-described first embodiment, the second correspondence table T2 may store the imaging conditions of the second medical image SM in the second modality device 20, but the present invention is not limited to this. That is, the second correspondence table T2 may store only the imaging conditions of the first medical image FM or the imaging conditions of the second medical image SM in the first modality device 10. In this case, since the processes of steps S41 to S45 of the conversion process shown in Fig. 11 are omitted, when it is determined in step S17 that the second medical image SM has been acquired as the medical data MD, the medical image processing device 50 may calculate the imaging conditions of the second medical image SM in the first modality device 10 in step S47.

[0121] Second Embodiment In the medical image processing apparatus 50 according to the first embodiment described above, the second medical image SM is acquired as medical data MD, and when the imaging conditions of the second medical image SM in the second modality device 20 are not stored, the imaging conditions of the second medical image SM in the first modality device 10 are calculated, but this is not limited thereto. In the medical image processing apparatus 50 according to the second embodiment, an image conversion method corresponding to the second medical image SM may be specified using a specific model without calculating the imaging conditions of the second medical image SM in the first modality device 10. Hereinafter, the differences from the first embodiment described above will be described.

[0122] FIG. 18 is a block diagram showing an example of the configuration of a medical image processing apparatus 50 according to the second embodiment, and corresponds to FIG. 3 in the first embodiment described above. As shown in FIG. 18, the medical image processing apparatus 50 according to this embodiment is configured by adding a second identification function 560 to the processing circuit 55 with respect to the medical image processing apparatus 50 according to the first embodiment described above. The second identification function 560 corresponds to the second identification unit according to this embodiment. In addition, the calculation function 555, the filtering function 556, and the selection function 557 may be omitted from the functions of the processing circuit 55 since they are not closely related to this embodiment. In addition, the configuration and functions other than the second identification function 560 are the same as those in FIG. 3 in the first embodiment described above, and therefore will not be described.

[0123] The second specifying function 560 uses the specific model to specify an image conversion method to be used in the image conversion processing of the second medical image SM. In this embodiment, the second specifying function 560 uses the specific model to specify an image conversion model to be used in the image conversion processing of the second medical image SM.

[0124] Fig. 19 is a flow chart for explaining the contents of the conversion process executed by the medical image processing apparatus 50 according to the second embodiment, and corresponds to Fig. 11 in the first embodiment described above. Note that the process up to step S45 shown in Fig. 19 is the same as Fig. 11 in the first embodiment described above, and therefore the explanation will be omitted.

[0125] 19, when it is determined in step S41 that the imaging conditions of the second medical image SM in the second modality device 20 are not stored (step S41: NO), the medical image processing device 50 identifies an image transformation model corresponding to the second medical image SM (step S101). The process of identifying the image transformation model corresponding to the second medical image SM is realized by the second identification function 560 in the processing circuitry 55. Specifically, the medical image processing device 50 inputs the second medical image SM to the identification model, thereby identifying the image transformation model used in the image conversion process of the second medical image SM.

[0126] Next, as shown in Fig. 19, the medical image processing apparatus 50 converts the second medical image SM into an input image IM using the specified image conversion model (step S103). This process of converting the second medical image SM into the input image IM is realized by the conversion function 552 in the processing circuitry 55. Specifically, the medical image processing apparatus 50 converts the second medical image SM into the input image IM by executing image conversion processing on the second medical image SM using the image conversion model specified in step S101. Note that the processes from step S61 onward after this step S103 are equivalent to those in Fig. 12 in the first embodiment described above, and therefore will not be described.

[0127] As described above, according to the medical image processing device 50 of the medical image processing system 1 of the second embodiment, when the medical image processing device 50 acquires a second medical image SM as medical data MD, it does not calculate the shooting conditions in the first modality device 10 of the second medical image SM, but uses a specific model to identify an image conversion model to be used in the image conversion processing of the second medical image SM, and converts the second medical image SM into an input image IM using the image conversion identified by the specific model.Therefore, an image captured by a different modality device can be applied to the trained model 210 without performing additional learning on the trained model 210.

[0128] In the above-described second embodiment, the second correspondence table T2 may store the imaging conditions of the second medical image SM in the second modality device 20, but the present invention is not limited to this. In other words, the second correspondence table T2 may store only the imaging conditions of the first medical image FM or the imaging conditions of the second medical image SM in the first modality device 10. In this case, since the processes of steps S41 to S45 of the conversion process shown in Fig. 19 are omitted, when it is determined in step S17 that the second medical image SM has been acquired as the medical data MD, the medical image processing device 50 may specify an image conversion model corresponding to the second medical image SM in step S101.

[0129] [Modifications of the first and second embodiments] The medical image-processing device 50 of the first and second embodiments described above can also return the adjusted usage ratio to the initial usage ratio before adjustment. Specifically, when the usage ratio is adjusted in step S67 of the conversion process, the medical image-processing device 50 can also return the adjusted usage ratio to the initial usage ratio based on the usage ratio information acquired in step S63. In order to return the adjusted usage ratio to the initial usage ratio before adjustment in this way, the medical image-processing device 50 outputs a reset button to the operation area OP2 for returning the usage ratio to the initial usage ratio. Then, the medical image-processing device 50 may return the usage ratio to the initial usage ratio before adjustment based on an input operation to the reset button by the user.

[0130] The medical image processing apparatus 50 of the first and second embodiments described above can also store an input image IM corresponding to the initial usage ratio before adjustment in the memory 51. Specifically, the medical image processing apparatus 50 can also store the input image IM output in step S61 of the conversion process, and the image conversion method information and usage ratio information acquired in step S63. The input image IM, the image conversion method information, and the usage ratio information may be stored automatically after the medical data MD is converted into the input image IM, or may be stored manually by the user by outputting a save button for saving the input image IM, the image conversion method information, and the usage ratio information in the operation area OP2.

[0131] In addition, when the input image IM and the image conversion method information and usage ratio information corresponding to the input image IM are stored in the memory 51, the medical image processing device 50 may read out the input image IM, the image conversion method information, and the usage ratio information stored in the memory 51 after the analysis result AR by the trained model 210 is output, adjust the usage ratio, and output the input image IM based on the adjusted usage ratio to the trained model 210 or the display 52.

[0132] The medical image processing apparatus 50 of the first and second embodiments described above can also be configured to additionally apply an image transformation model to the input image IM converted by applying the image transformation model. Specifically, the medical image processing apparatus 50 may apply the image transformation model selected for additional application to the input image IM output in step S61 based on an input operation for selecting the image transformation model to be additionally applied. More specifically, for example, when the contrast transformation model A is applied to the input image IM output in step S61, the medical image processing apparatus 50 may apply another image transformation model such as the contrast transformation model B to the input image IM to which the contrast transformation model A is applied. In this case, the medical image processing apparatus 50 outputs a selection button for selecting the image transformation model to be additionally applied to the operation area OP2. Then, the medical image processing apparatus 50 may additionally apply the selected image transformation model to the input image IM based on an input operation on the selection button by the user, and output the input image IM to which the image transformation model has been additionally applied to the display 52.

[0133] The medical image processing device 50 of the first and second embodiments described above may preset a pattern set that is a set including a plurality of patterns for applying an image conversion model to the first medical image FM or the second medical image SM, and output an input image IM to the display 52 in which the image conversion model for each pattern is applied to the first medical image FM or the second medical image SM based on the preset pattern set. The medical image processing device 50 may also store in the memory 51 an input image IM in which the image conversion model for each pattern is applied to the first medical image FM or the second medical image SM based on the preset pattern set. Specifically, for example, the medical image processing device 50 may preset a pattern for applying a contrast conversion model A, a pattern for applying a noise conversion model A, and a pattern for applying a high image quality conversion model A as the pattern set, and convert the first medical image FM or the second medical image SM acquired by the acquisition function 551 into three input images IM in which the contrast conversion model A, the noise conversion model A, and the high image quality conversion model A are applied, respectively.

[0134] In the above example, the pattern set includes three patterns: a pattern that applies the contrast conversion model A, a pattern that applies the noise conversion model A, and a pattern that applies the high image quality conversion model A. However, the number of patterns included in the pattern set is not limited to three. In other words, the number of patterns included in the pattern set is arbitrary, and may include one or two patterns, or may include four or more patterns.

[0135] In the conversion process executed in the medical image processing device 50 of the first and second embodiments described above, the case where the converted input image IM is output to the display 52 in steps S35, S39, S53, and S59 has been considered, but is not limited to this. That is, in the conversion process, the medical image processing device 50 may output the converted input image IM to the trained model 210. In this case, since the processes from step S61 to step S73 in the conversion process are omitted, when the medical image processing device 50 converts the input image IM to the input image IM in steps S35, S39, S53, and S59, it may output the input image IM to the trained model 210 in step S75.

[0136] The term "processor" used in the above description means a circuit such as a central processing unit (CPU), a graphics processing unit (GPU), an application specific integrated circuit (ASIC), a programmable logic device (e.g., a simple programmable logic device (SPLD), a complex programmable logic device (CPLD), and a field programmable gate array (FPGA)). The processor realizes its function by reading and executing a program stored in the memory 51. Instead of storing a program in the memory 51, the processor may be configured to directly incorporate the program in its circuit. In this case, the processor realizes its function by reading and executing the program incorporated in its circuit. The processor is not limited to being configured as a single processor circuit, and may be configured as a single processor by combining multiple independent circuits to realize its function. Furthermore, the multiple components in FIG. 3 may be integrated into a single processor to realize its function.

[0137] Although several embodiments have been described above, these embodiments are presented only as examples and are not intended to limit the scope of the invention. The novel device and method described in this specification can be implemented in various other forms. In addition, various omissions, substitutions, and modifications can be made to the forms of the device and method described in this specification without departing from the spirit of the invention. The appended claims and their equivalents are intended to include such forms and modifications that fall within the scope and spirit of the invention. [Explanation of symbols]

[0138] 1...medical image processing system, 10...first modality device, 20...second modality device, 30...third modality device, 40...medical data server, 50...medical image processing device, 51...memory, 52...display, 53...input interface, 54...communication interface, 55...processing circuit, 210...trained model, 551...acquisition function, 552...conversion function, 553...output function, 554...first identification function, 555...calculation function, 556...filtering function, 557...selection function, 558...adjustment function, 559...registration function, 560...second identification function

Claims

1. An acquisition unit for acquiring medical data; A conversion unit that converts the medical data acquired by the acquisition unit into an input image that matches an input of a trained model that outputs an image analysis result; A medical image processing device comprising:

2. The medical image processing apparatus according to claim 1 , further comprising an output unit that outputs the input image converted by the conversion unit.

3. a first storage unit in which image conversion conditions, which are conditions for converting raw data collected in a first modality device into the input image, are stored in association with imaging conditions of the raw data; 3. The medical image processing device according to claim 2, wherein when the acquisition unit acquires the raw data as the medical data, the conversion unit acquires the image conversion conditions corresponding to the shooting conditions of the raw data, and converts the raw data into the input image by performing image conversion processing on the raw data using the acquired image conversion conditions.

4. a second storage unit in which an image conversion method for converting a first medical image captured by a first modality device into the input image is stored in association with an imaging condition of the first medical image; 3. The medical image processing device according to claim 2, wherein when the acquisition unit acquires the first medical image as the medical data, the conversion unit converts the first medical image into the input image by performing an image conversion process on the first medical image using the image conversion method stored in the second storage unit.

5. The conversion unit is determining whether or not the imaging conditions of the first medical image acquired by the acquisition unit are stored in the second storage unit; 5. The medical image processing device of claim 4, wherein when the shooting conditions of the first medical image are stored in the second storage unit, the image conversion method corresponding to the shooting conditions of the first medical image is acquired, and the first medical image is converted into the input image by performing image conversion processing using the acquired image conversion method on the first medical image.

6. A first identification unit is further provided that identifies the image conversion method to be used in the image conversion processing of the first medical image by using an identification model that is a model that identifies the image conversion method to be used in the image conversion processing of the medical image based on the medical image, When the imaging conditions of the first medical image are not stored in the second storage unit, the first identification unit inputs the first medical image to the identification model to identify the image conversion method to be used in the image conversion process of the first medical image; The medical image processing device according to claim 5 , wherein the conversion unit converts the first medical image into the input image by performing an image conversion process on the first medical image using the image conversion method identified by a first identification unit.

7. 5. The medical image processing device according to claim 4, wherein when the acquisition unit acquires a second medical image captured by a second modality device different from the first modality device as the medical data, the conversion unit converts the second medical image into the input image by performing image conversion processing on the second medical image using the image conversion method stored in the second storage unit.

8. a calculation unit that calculates an imaging condition of the first modality device of the second medical image based on a second medical image captured by a second modality device different from the first modality device, The conversion unit is determining whether or not the imaging conditions of the second medical image in the first modality device calculated by the calculation unit are stored in the second storage unit; When the imaging conditions of the second medical image in the first modality device are stored in the second storage unit, the image conversion method corresponding to the imaging conditions of the second medical image in the first modality device is acquired; converting the first medical image into the input image by performing an image conversion process on the second medical image using the acquired image conversion method; The medical image processing device according to claim 7 .

9. a filtering unit that filters the imaging conditions of the first medical image stored in the second storage unit based on the imaging conditions of the first modality device of the second medical image calculated by the calculation unit; A selection unit that selects the image conversion method to be used in the image conversion processing of the second medical image from the image conversion methods associated with the imaging conditions of the first medical image filtered by the filtering unit based on the second medical image and the input image, 9. The medical image processing device according to claim 8, wherein, when the shooting conditions of the second medical image in the first modality device are not stored in the second storage unit, the conversion unit converts the second medical image into the input image by performing image conversion processing on the second medical image using the image conversion method selected by the selection unit.

10. A second identification unit is further provided that identifies the image conversion method to be used in the image conversion processing of the second medical image by using an identification model that is a model that identifies the image conversion method to be used in the image conversion processing of the medical image based on the medical image, The second identification unit inputs the second medical image to the identification model to identify the image conversion method used in the image conversion process of the second medical image; The medical image processing device according to claim 7 , wherein the conversion unit converts the second medical image into the input image by performing an image conversion process on the second medical image using the image conversion method identified by the second identification unit.

11. The medical image processing apparatus according to claim 5 , wherein the image conversion method is an image conversion model generated by machine learning based on the first medical image and an imaging condition of the first medical image.

12. The medical image processing apparatus according to claim 5 , wherein the image conversion method is a rule-based conversion formula generated based on the first medical image and an imaging condition of the first medical image.

13. The medical image processing apparatus according to claim 2 , wherein the output unit outputs the input image to the trained model.

14. 13. The medical image processing device according to claim 4, wherein the output unit outputs, together with the analysis result, to a display unit, image conversion method information, which is information regarding the image conversion method used when converting the image to the input image, and usage ratio information, which is information regarding the usage ratio of the image conversion method used when converting the image to the input image.

15. an adjustment unit that receives an input operation for adjusting a usage ratio of the image conversion method from a user and adjusts the usage ratio based on the received input operation, The medical image processing apparatus according to claim 14 , wherein the output unit outputs the input image based on the adjusted usage ratio to the display unit.

16. The medical image processing apparatus according to claim 14 , further comprising a registration unit that registers the image conversion method information and the imaging conditions in a second storage unit in association with each other.

17. The image conversion method information includes used image ratio information, which is information regarding the image ratio used when generating the image conversion method, The medical image processing apparatus according to claim 14 , wherein the output unit outputs the used image ratio information to the display unit.

18. a medical data server for storing medical data; an acquisition unit that acquires the medical data from the medical data server; A conversion unit that converts the medical data acquired by the acquisition unit into an input image that matches an input of a trained model that outputs an image analysis result, A medical image processing device; A medical image processing system comprising:

19. acquiring medical data; A step of converting the acquired medical data into an input image that matches an input of a trained model that outputs an image analysis result; A program for causing a medical image processing device to execute the above.

Citation Information

Patent Citations

  • Image processing device, medical diagnostic system, learned model, image processing method, and program

    JP2020039490A

  • Medical information processing device and medical image diagnostic system

    JP2022035508A

  • Information processing device and information processing method

    WO2023281738A1

Cited By

  • Medical image processing device, model learning method, and program

    JP2026122410A