Diagnosis support device, diagnosis support method, and diagnosis support program

Abstract

This diagnosis support device is provided with: a first discrimination unit that, on the basis of first input information including at least one of information indicating an abnormal site detected from an image for diagnosis of a patient to be diagnosed and medical information of the patient, outputs a first discrimination result relating to the medical condition of the patient; a second discrimination unit that, on the basis of second input information including at least one of information indicating the abnormal site detected from the image for diagnosis and the medical information, outputs a second discrimination result relating to the medical condition of the patient; and a discussion unit that, when the first discrimination result and the second discrimination result are different, outputs a discussion result indicating the content of a discussion between the standpoint of the first discrimination result and the standpoint of the second discrimination result.

Description

Diagnostic support device, diagnostic support method, and diagnostic support program

[0001] The present disclosure relates to a diagnostic support device, a diagnostic support method, and a diagnostic support program.

[0002] Technologies for assisting diagnosis are known. One example of such a technology is described in Patent Literature 1. Patent Literature 1 describes a technology that enables a case conference to be held among multiple doctors in remote locations, in which messages posted to a virtual conference generated for diagnostic data are received from terminal devices of users participating in the virtual conference, the posted messages are transmitted to terminal devices other than the terminal devices, the posted messages are analyzed, and information related to the diagnostic data obtained based on the analysis results is generated, and the information related to the diagnostic data is transmitted to the terminal devices.

[0003] International Publication No. 2019 / 102749

[0004] However, there are cases where multiple opinions are possible regarding a patient's condition, and it would be ideal if there was support for determining which opinion is appropriate in such cases. The technology described in Patent Document 1 has a problem in that it cannot support the determination of which opinion is appropriate when there are multiple opinions regarding a patient's condition. In addition, there are facilities where it is difficult for multiple doctors to hold discussions due to a shortage of doctors.

[0005] The present disclosure has been made in view of the above-mentioned problems, and an exemplary purpose thereof is to provide a technology that can more appropriately assist in the diagnosis of a patient who is the subject of diagnosis.

[0006] A diagnostic support device according to an exemplary aspect of the present disclosure comprises a first discrimination means for outputting a first discrimination result regarding the condition of a patient based on first input information including at least one of information indicating an abnormal location detected from a diagnostic target image of the patient and medical information of the patient; a second discrimination means for outputting a second discrimination result regarding the condition of the patient based on second input information including at least one of information indicating an abnormal location detected from the diagnostic target image and the medical information; and a discussion means for outputting a discussion result indicating the content of a discussion between the standpoint of the first discrimination result and the standpoint of the second discrimination result when the first discrimination result and the second discrimination result differ.

[0007] A diagnostic support method according to an exemplary aspect of the present disclosure includes a first discrimination process in which at least one processor outputs a first discrimination result regarding the condition of a patient based on first input information including at least one of information indicating an abnormal location detected from a diagnostic target image of the patient and medical information of the patient; a second discrimination process in which the at least one processor outputs a second discrimination result regarding the condition of the patient based on second input information including at least one of information indicating an abnormal location detected from the diagnostic target image and the medical information; and a discussion process in which the at least one processor outputs a discussion result indicating the content of a discussion from the perspective of the first discrimination result and the perspective of the second discrimination result if the first discrimination result and the second discrimination result differ.

[0008] A diagnostic assistance program according to an exemplary aspect of the present disclosure is a program for causing a computer to function as a diagnostic assistance program, and causes the computer to function as: a first discrimination means for outputting a first discrimination result regarding the condition of a patient based on first input information including at least one of information indicating an abnormal location detected from a diagnostic target image of the patient and medical information of the patient; a second discrimination means for outputting a second discrimination result regarding the condition of the patient based on second input information including at least one of information indicating an abnormal location detected from the diagnostic target image and the medical information; and a discussion means for outputting a discussion result indicating the content of a discussion from the perspective of the first discrimination result and the perspective of the second discrimination result when the first discrimination result and the second discrimination result differ.

[0009] According to an exemplary aspect of the present disclosure, an exemplary effect is achieved in that a technology can be provided that more appropriately supports the diagnosis of a patient who is the subject of diagnosis.

[0010] FIG. 1 is a block diagram showing the configuration of a diagnosis support device according to the present disclosure. FIG. 2 is a flow diagram showing the flow of a diagnosis support method according to the present disclosure. FIG. 3 is a block diagram showing the configuration of an information processing device according to the present disclosure. FIG. 4 is a diagram showing a specific example of a diagnostic target image and information showing an abnormality detected from the diagnostic target image according to the present disclosure. FIG. 5 is a diagram showing a specific example of a discrimination result that is the output of a discrimination model according to the present disclosure. FIG. 6 is a diagram showing a specific example of a discrimination result obtained by inputting each of a plurality of pieces of input information according to the present disclosure into a discrimination model. FIG. 7 is a diagram showing a specific example of a discussion result according to the present disclosure. FIG. 8 is a flow diagram showing an example of the flow of a diagnosis support method according to the present disclosure. FIG. 9 is a block diagram showing the configuration of an information processing device according to the present disclosure. FIG. 10 is a block diagram showing the configuration of an information processing device according to the present disclosure. FIG. 11 is a diagram showing a specific example of a plurality of captions generated by a caption generation unit according to the present disclosure. FIG. 12 is a block diagram showing the configuration of an information processing device according to the present disclosure. FIG. 13 is a diagram showing a specific example of a discrimination result according to the present disclosure. FIG. 14 is a diagram showing another example of a discrimination method according to the present disclosure. FIG. 15 is a block diagram showing the configuration of a computer that functions as a diagnosis support device and an information processing device according to the present disclosure.

[0011] The following are examples of embodiments of the present invention. However, the present invention is not limited to the exemplary embodiments shown below, and various modifications are possible within the scope of the claims. For example, embodiments obtained by appropriately combining the technical means employed in the exemplary embodiments shown below may also be included in the scope of the present invention. Furthermore, embodiments obtained by appropriately omitting some of the technical means employed in the exemplary embodiments shown below may also be included in the scope of the present invention. Furthermore, the effects mentioned in the exemplary embodiments shown below are examples of effects expected in the exemplary embodiments, and do not define the scope of the present invention. In other words, embodiments that do not exhibit the effects mentioned in the exemplary embodiments shown below may also be included in the scope of the present invention.

[0012] [First Exemplary Embodiment] A first exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. This exemplary embodiment is a basic form for each of the exemplary embodiments described below. Note that the scope of application of each technical means employed in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technical means employed in this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical obstacles arise. Furthermore, each technical means shown in the drawings referenced to explain this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical obstacles arise.

[0013] (Configuration of the Diagnostic Support Device) The configuration of the diagnostic support device 1 will be described with reference to FIG. 1. FIG. 1 is a block diagram showing the configuration of the diagnostic support device 1. As shown in FIG. 1, the diagnostic support device 1 includes a first discrimination unit 11, a second discrimination unit 12, and a discussion unit 13. The first discrimination unit 11 outputs a first discrimination result regarding the condition of a patient based on first input information including at least one of information indicating an abnormality detected from a diagnostic target image of the patient and medical information of the patient. The second discrimination unit 12 outputs a second discrimination result regarding the condition of the patient based on second input information including at least one of information indicating an abnormality detected from the diagnostic target image and the medical information. If the first discrimination result and the second discrimination result differ, the discussion unit 13 outputs a discussion result indicating the content of a discussion between the perspectives of the first discrimination result and the second discrimination result.

[0014] (Effects of the Diagnostic Support Device) As described above, the diagnostic support device 1 includes a first discrimination unit 11 that outputs a first discrimination result regarding the patient's condition based on first input information including at least one of information indicating an abnormality detected from a diagnostic target image of the patient and medical information of the patient, a second discrimination unit 12 that outputs a second discrimination result regarding the patient's condition based on second input information including at least one of information indicating an abnormality detected from the diagnostic target image and medical information of the patient, and a discussion unit 13 that outputs a discussion result indicating the content of a discussion between the perspectives of the first discrimination result and the second discrimination result when the first discrimination result and the second discrimination result differ. Therefore, the diagnostic support device 1 has the effect of more appropriately supporting the diagnosis of the patient.

[0015] (Flow of Diagnostic Support Method) The flow of the diagnostic support method S1 will be described with reference to FIG. 2. FIG. 2 is a flow chart showing the flow of the diagnostic support method S1. As shown in FIG. 2, the diagnostic support method S1 includes a first discrimination process S11, a second discrimination process S12, and a discussion process S13. In the first discrimination process S11, at least one processor outputs a first discrimination result regarding the condition of a patient based on first input information including at least one of information indicating an abnormality detected from a diagnostic target image of the patient and medical information of the patient. In the second discrimination process S12, at least one processor outputs a second discrimination result regarding the condition of the patient based on second input information including at least one of information indicating an abnormality detected from the diagnostic target image and medical information. In the discussion process S13, if the first discrimination result and the second discrimination result differ, at least one processor outputs a discussion result indicating the content of a discussion between the perspectives of the first discrimination result and the second discrimination result.

[0016] (Effects of the Diagnostic Support Method) As described above, the diagnostic support method S1 includes a first discrimination process in which at least one processor outputs a first discrimination result regarding the patient's condition based on first input information including at least one of information indicating an abnormality detected from a diagnostic target image of the patient and medical information of the patient, a second discrimination process in which the at least one processor outputs a second discrimination result regarding the patient's condition based on second input information including at least one of information indicating an abnormality detected from the diagnostic target image and medical information of the patient, and a discussion process in which the at least one processor outputs a discussion result indicating the content of a discussion between the perspectives of the first discrimination result and the second discrimination result if the first discrimination result and the second discrimination result differ. Therefore, the diagnostic support method S1 has the effect of more appropriately supporting the diagnosis of the patient.

[0017] Second Exemplary Embodiment A second exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. Components having the same functions as those described in the above exemplary embodiment will be denoted by the same reference numerals, and their description will be omitted as appropriate. The scope of application of each technical means employed in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technical means employed in this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs. Furthermore, each technical means shown in each drawing referenced to describe this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs.

[0018] (Configuration of information processing device) The information processing device 1A is a device that outputs information to support diagnosis. The information output by the information processing device 1A is used, for example, to support the decision-making of medical professionals. The information processing device 1A is an example of a diagnosis support device according to the present disclosure. The configuration of the information processing device 1A will be described with reference to FIG. 3. FIG. 3 is a block diagram showing the configuration of the information processing device 1A. The information processing device 1A includes a control unit 10A, a memory unit 20A, a communication unit 30A, an input unit 40A, and an output unit 50A.

[0019] (Communication Unit) The communication unit 30A communicates with devices external to the information processing device 1A via a communication line. While the specific configuration of the communication line does not limit the present exemplary embodiment, examples of the communication line include a wireless local area network (LAN), a wired LAN, a wide area network (WAN), a public line network, a mobile data communication network, or a combination thereof. The communication unit 30A transmits data supplied from the control unit 10A to other devices, and supplies data received from other devices to the control unit 10A.

[0020] (Input Unit) The input unit 40A is configured to receive input to the information processing device 1A, and includes, for example, input devices such as a keyboard, a mouse, a touch panel, a camera, a microphone, etc. The input unit 40A may also be configured to receive data from the input devices via an interface such as a USB (Universal Serial Bus).

[0021] (Output Unit) The output unit 50A is a component for performing output from the information processing device 1A, and includes, for example, output devices such as a display, a printer, a touch panel, a speaker, etc. The output unit 50A may be configured to include, for example, an interface such as a USB, and to output data to the output device via the interface.

[0022] (Memory Unit) The memory unit 20A stores various types of information referenced by the control unit 10A. Examples of such information include medical information 201 and diagnostic target images 202. The medical information 201 is information related to the medical care of a patient. Examples of the medical information 201 include chart information. Examples of the medical information include the patient's age, gender, chief complaint, medical history, family medical history, smoking amount, etc. The diagnostic target images 202 are images of the patient used for diagnosis. Examples of diagnostic target images include at least one of X-ray images, endoscopic images, pathological images, MRI images, and CT images. The memory unit 20A stores medical information 201 and diagnostic target images 202 for each of a plurality of patients.

[0023] (Control unit) The control unit 10A includes a lesion detection unit 11A, a caption generation unit 12A, a classification unit 13A, a discussion unit 14A, a presentation unit 15A, a search unit 16A, and a search result output unit 17A. The lesion detection unit 11A is an example of a detection means according to the present disclosure. The caption generation unit 12A is an example of a caption generation means according to the present disclosure. The classification unit 13A is an example of a first classification means and a second classification means according to the present disclosure. The discussion unit 14A is an example of a discussion means according to the present disclosure. The presentation unit 15A is an example of a presentation means according to the present disclosure. The search unit 16A is an example of a search means according to the present disclosure. The search result output unit 17A is an example of a search result output means according to the present disclosure.

[0024] (Lesion Detection Unit) The lesion detection unit 11A acquires a diagnostic target image 202 of a patient who is the subject of diagnosis and detects abnormal areas from the diagnostic target image 202. As an example, the lesion detection unit 11A may acquire the diagnostic target image 202 by reading the diagnostic target image 202 from a storage location (which may be a storage device within the information processing device 1A or a storage device external to the information processing device 1A) specified by the user of the information processing device 1A. Alternatively, the lesion detection unit 11A may acquire the diagnostic target image 202 by receiving the diagnostic target image 202 from another device via the communication unit 30A. Alternatively, the lesion detection unit 11A may acquire the diagnostic target image 202 input to the input unit 40A.

[0025] As an example, the lesion detection unit 11A detects abnormal areas from the diagnostic target image 202 using a detection model that detects lesions from an image. Hereinafter, abnormal areas in the diagnostic target image 202 are also referred to as "lesion areas." The detection model may be stored in the storage unit 20A of the information processing device 1A, or may be stored in a device other than the information processing device 1A. Note that "the detection model is stored in the storage unit 20A" means that parameters that define the detection model are stored in the storage unit 20A.

[0026] The detection model is, for example, a trained model generated by machine learning. Examples of the detection model include, but are not limited to, Regional CNN (R-CNN), You Only Look Once (YOLO), Single Shot MultiBox Detector (SDD), and End-to-End Object Detection with Transformers (DETR).

[0027] An example of the input of the detection model is the diagnostic target image 202. The output of the detection model includes information indicating an abnormal location (lesion). An example of the information indicating an abnormal location is information indicating the area of ​​the lesion (coordinate information, etc.). The information indicating an abnormal location may also include data representing an image in which the lesion location is marked in the diagnostic target image 202.

[0028] When the detection model is stored in a device other than information processing device 1A, lesion detection unit 11A, for example, inputs diagnostic object image 202 to the detection model by transmitting diagnostic object image 202 to the device storing the detection model via communication unit 30A. In this case, lesion detection unit 11A acquires information output by the detection model by receiving it from the device via communication unit 30A.

[0029] 4 is a diagram showing a specific example of a diagnostic target image 202 and information indicating an abnormality detected from the diagnostic target image. In the example of FIG. 4, lesion detection unit 11A inputs diagnostic target image 202 to lesion detection model M1. In addition, lesion detection model M1 outputs image 203 in which a lesion location is marked in diagnostic target image 202 as information indicating the abnormality.

[0030] (Caption Generation Unit) The caption generation unit 12A generates a caption indicating an abnormal location detected by the lesion detection unit 11A. Here, the caption is a sentence that explains the abnormal location in the diagnostic target image 202, such as a sentence such as "A white nodule near XX in the upper left lobe." The caption is an example of information indicating an abnormal location detected in the diagnostic target image of a patient.

[0031] As an example, the caption generation unit 12A uses a caption generation model that generates a caption from an image to generate a caption from the information indicating the abnormal location and the diagnostic target image 202. In this case, the caption generation model may be stored in the storage unit 20A of the information processing device 1A, or may be stored in a device other than the information processing device 1A. Note that "the caption generation model is stored in the storage unit 20A" means that parameters that define the caption generation model are stored in the storage unit 20A.

[0032] The caption generation model is, for example, a trained model generated by machine learning. Examples of the caption generation model include, but are not limited to, an existing generation AI such as CoCa (Contrastive Captioner) that has been fine-tuned using medical images.

[0033] As an example, the input of the caption generation model includes the diagnostic target image 202 and information indicating the abnormal location, which is the output of the detection model. As another example, the input of the caption generation model may be an image output of the detection model, i.e., an image in which the lesion location is marked in the diagnostic target image 202. Furthermore, if the output of the detection model is coordinate information indicating the position of the abnormal location, the caption generation unit 12A may generate a superimposed image in which a mark image is superimposed on the diagnostic target image 202 at a position indicated by the position information, and use the generated superimposed image as input to the caption generation model. The output of the caption generation model is a caption.

[0034] When the caption generation model is stored in a device other than the information processing device 1A, the caption generation unit 12A, for example, inputs input information including the diagnostic target image 202 and information that is the output of the detection model to the device that stores the caption generation model via the communication unit 30A, thereby inputting the input information to the caption generation model. In this case, the caption generation unit 12A obtains the caption output by the caption generation model by receiving it from the device via the communication unit 30A.

[0035] 4, the caption generation unit 12A inputs the image 203 output from the lesion detection model M1 to the caption generation model M2, which then outputs a caption 204 for the image 203.

[0036] (Discrimination unit) The discrimination unit 13A generates a discrimination result regarding the patient's condition. As an example, the discrimination unit 13A generates the discrimination result using a discrimination model generated by machine learning. The discrimination model may be stored in the memory unit 20A of the information processing device 1A, or may be stored in a device other than the information processing device 1A. Note that the discrimination model being stored in the memory unit 20A means that parameters defining the discrimination model are stored in the memory unit 20A.

[0037] The discrimination model is, for example, a trained model generated by machine learning. Examples of the discrimination model include, but are not limited to, trained models generated by supervised learning using techniques such as neural networks, or models in which generative AI such as ChatGPT (Chat Generative Pre-trained Transformer) and GPT-4 (Generative Pre-trained Transformer 4) is fine-tuned using medical information.

[0038] The input information input to the discrimination model includes, for example, a prompt. Here, the prompt includes at least one of the patient's medical information 201 and a caption generated by the caption generation unit 12A. The caption included in the prompt is an example of information indicating an abnormal area detected from the diagnostic target image 202. The input information may also include at least one of the diagnostic target image 202 and information (coordinate information, image data, etc.) indicating the abnormal area detected from the diagnostic target image 202. The output of the discrimination model is information indicating a discrimination result. The discrimination result may be, for example, information indicating a "metastatic tumor," "inflammatory nodule," or "pneumothorax."

[0039] When the classification model is stored in a device other than the information processing device 1A, the classification unit 13A inputs the input information to the detection model by, for example, transmitting the input information to the device storing the classification model via the communication unit 30A. In this case, the classification unit 13A acquires the information output by the classification model by receiving it from the device via the communication unit 30A.

[0040] The classification unit 13A acquires medical information 201 of a patient to be diagnosed, generates a prompt including the acquired medical information 201 and the caption generated by the caption generation unit 12A, and inputs input information including the generated prompt to the classification model. For example, the classification unit 13A may acquire the medical information 201 by reading the medical information 201 from a storage location (which may be a storage device within the information processing device 1A or a storage device external to the information processing device 1A) specified by the user of the information processing device 1A. Alternatively, the classification unit 13A may acquire the medical information 201 by receiving the medical information 201 from another device via the communication unit 30A. Alternatively, the classification unit 13A may acquire the medical information 201 input to the input unit 40A.

[0041] 5 is a diagram showing a specific example of a discrimination result that is the output of the discrimination model M3. In the example of FIG. 5, the discrimination unit 13A inputs a prompt 205 to the discrimination model M3. The prompt 205 includes medical information 201 of the patient to be diagnosed and a caption generated by the caption generation unit 12A. Furthermore, the discrimination model M3 outputs "metastatic tumor" as the discrimination result.

[0042] (Generation of Multiple Input Information) The classification unit 13A generates multiple pieces of input information and outputs a classification result regarding the patient's condition based on each of the generated multiple pieces of input information. In other words, the classification unit 13A outputs a first classification result regarding the patient's condition based on first input information including at least one of information indicating an abnormality detected from a diagnostic target image of the patient who is the diagnostic target and medical information of the patient, and outputs a second classification result regarding the patient's condition based on second input information including at least one of information indicating an abnormality detected from the diagnostic target image and medical information. Furthermore, when there are three or more pieces of input information, the classification unit 13A can also be said to output a third classification result regarding the patient's condition based on third input information including at least one of information indicating an abnormality detected from the diagnostic target image and medical information.

[0043] In this exemplary embodiment, the second input information is information different from the first input information. For example, the second input information may include some of the information included in the first input information, but not include other parts of the information included in the first input information. More specifically, for example, the prompts included in the second input information may include some of the prompts included in the first input information, but not include other parts of the information included in the first input information. For another example, the first input information may include a diagnostic target image, and the second input information may not include the diagnostic target image.

[0044] FIG. 6 illustrates a specific example of a discrimination result obtained by inputting each of multiple pieces of input information into the discrimination model M3. In the example of FIG. 6, the discrimination unit 13A inputs each of the pieces of input information 205a to 205f into the discrimination model M3. The input information 205a to 205f are different from each other. More specifically, in the input information 205a to 205f, one different item from the multiple items included in the prompt has been deleted. For example, the input information 205a does not include the text "48 years old, male" included in the other pieces of input information 205b to 205f. Furthermore, the input information 205b does not include the text "medical history: colon cancer" included in the other pieces of input information 205a, 205c to 205f. Furthermore, the input information 205f does not include the caption (the caption generated by the caption generation unit 12A) included in the other pieces of input information 205a to 205e.

[0045] In the example of Figure 6, among the discrimination results 206a to 206f obtained by inputting input information 205a to 205f into discrimination model M3, discrimination results 206a, 206c, 206d, and 206e are information indicating "metastatic tumor." On the other hand, discrimination result 206b is information indicating "inflammatory nodule," and discrimination result 206f is information indicating "pneumothorax." In other words, discrimination results 206b and 206f are different from discrimination result 206a. In this way, by varying the input to discrimination model M3, multiple different discrimination results can be obtained.

[0046] (Discussion Section) When the first discrimination result and the second discrimination result differ, the discussion section 14A outputs a discussion result indicating the content of the discussion from the standpoint of the first discrimination result and the standpoint of the second discrimination result. The discussion result is, for example, information used to support decision-making regarding the diagnosis of a patient's condition. Furthermore, when three or more different discrimination results are obtained, the discussion section 14A can also be said to output a discussion result based on the first discrimination result, the second discrimination result, and the third discrimination result.

[0047] As an example, the discussion unit 14A outputs the discussion results using a generative model that generates text. In this case, the generative model may be stored in the storage unit 20A of the information processing device 1A, or may be stored in a device other than the information processing device 1A. Note that storing the generative model in the storage unit 20A means that parameters that define the generative model are stored in the storage unit 20A.

[0048] An example of the generative model is a trained model generated by machine learning. Examples of the generative model include, but are not limited to, models in which a generative AI such as ChatGPT or GPT-4 is fine-tuned using medical information. The generative model may be a caption generation model used by the caption generation unit 12A.

[0049] The input information input to the generative model includes, for example, a prompt generated by the caption generation unit 12A and a sentence instructing a discussion from each perspective of multiple discrimination results. The input information may also include a diagnostic target image 202. The input information may also include information indicating abnormalities generated by the lesion detection unit 11A (coordinate information indicating abnormalities, image data with abnormalities marked, etc.). The output of the generative model includes a discussion result. For example, the discussion result may be in text format or may be an image representing the discussion result.

[0050] When the generative model is stored in a device other than the information processing device 1A, the discussion unit 14A inputs the input information to the generative model by, for example, transmitting the input information to be input to the generative model to the device storing the generative model via the communication unit 30A. In this case, the discussion unit 14A obtains the discussion results output by the generative model by receiving them from the device via the communication unit 30A.

[0051] 7 is a diagram showing a specific example of a discussion result. In the example of FIG. 7, the discussion unit 14A generates input information 401 including a prompt generated by the caption generation unit 12A and text instructing discussion from the perspectives corresponding to each of a plurality of differentiation results. In the example of FIG. 7, the input information 401 includes the text "Please discuss the patient's diagnosis results from the perspectives of three doctors," as well as the prompt generated by the caption generation unit 12A, medical information of the patient to be diagnosed, and the text "Doctor 1 - perspective of diagnosing metastatic tumor," "Doctor 2 - perspective of diagnosing inflammatory nodule," and "Doctor 3 - perspective of diagnosing pneumothorax."

[0052] The discussion unit 14A inputs the generated input information 401 into the generative model. Furthermore, the generative model outputs the results of the discussion conducted from the perspectives of the three doctors. The discussion unit 14A outputs the discussion results. For example, the discussion unit 14A may output the discussion results by writing the discussion results to a storage destination (which may be a storage device within the information processing device 1A or a storage device external to the information processing device 1A) designated by the user of the information processing device 1A. Furthermore, the discussion unit 14A may output the discussion results by transmitting the discussion results via the communication unit 30A, or may output the discussion results to an output device such as a display.

[0053] (Presentation Unit) When the classification unit 13A obtains a plurality of different classification results, the presentation unit 15A presents information indicating the difference between the first input information and the second input information. For example, the presentation unit 15A presents the information by outputting the information to an output device such as a display. Alternatively, the presentation unit 15A may present the information by transmitting the information to an output device connected via the communication unit 30A.

[0054] For example, the presenting unit 15A may display the prompt included in the first input information and the prompt included in the second input information side by side on the display device. The presenting unit 15A may also output, among the information included in the prompt generated by the discriminating unit 13A, information that is not included in the first input information as information indicating the difference, and may also output, as information indicating the difference, information that is not included in the second input information.

[0055] (Search Unit) The search unit 16A extracts multiple points of discussion from the discussion results of the discussion unit 14A and searches a database for literature related to the extracted points of discussion. As an example, the search unit 16A extracts the points of discussion using a generative model that generates text. In this case, the generative model may be stored in the storage unit 20A of the information processing device 1A, or may be stored in a device other than the information processing device 1A. Note that "the generative model is stored in the storage unit 20A" means that parameters that define the generative model are stored in the storage unit 20A.

[0056] An example of the generative model is a trained model generated by machine learning. Examples of the generative model include, but are not limited to, models in which generative AI such as ChatGPT and GPT-4 is fine-tuned using medical information. The generative model may be a generative model used by the discussion unit 14A, or a caption generation model used by the caption generation unit 12A.

[0057] The input of the generative model includes, for example, the discussion results generated by the discussion unit 14A. The output of the generative model includes the extraction results of one or more points of discussion. For example, the extraction results may be in text format or may be an image representing the discussion results.

[0058] When the generative model is stored in a device other than the information processing device 1A, the search unit 16A, for example, inputs the discussion results to the generative model by transmitting the discussion results to the device storing the generative model via the communication unit 30A. In this case, the search unit 16A obtains the extraction results output by the generative model by receiving them from the device via the communication unit 30A.

[0059] As an example, the search unit 16A searches a predetermined database using the extracted issue or a word contained in the issue as a search key. The database may be stored in the storage unit 20A of the information processing device 1A, or may be stored in a device other than the information processing device 1A. When the database is stored in a device other than the information processing device 1A, as an example, the search unit 16A transmits the search key to a device that stores the generative model via the communication unit 30A. The device that receives the search key searches the database for documents that match the received search key and transmits the documents that match the search key to the information processing device 1A. The search unit 16A receives the documents from the above device.

[0060] The search unit 16A may also output one or more points of discussion extracted from the discussion results. For example, the search unit 16A may output the points of discussion by writing them to a storage location (which may be a storage device within the information processing device 1A or a storage device external to the information processing device 1A) designated by the user of the information processing device 1A. The search unit 16A may also output the points of discussion by transmitting them via the communication unit 30A, or may output them to an output device such as a display.

[0061] (Search result output unit) The search result output unit 17A outputs the search results obtained by the search unit 16A. As an example, the search result output unit 17A may output the search results by writing the search results to a storage destination (which may be a storage device within the information processing device 1A or a storage device external to the information processing device 1A) designated by the user of the information processing device 1A. Furthermore, the search result output unit 17A may output the search results by transmitting the search results via the communication unit 30A, or may output the search results to an output device such as a display.

[0062] The search result output unit 17A may also select one or more documents from the plurality of documents searched by the search unit 16A and output the selected documents. In this case, the search result output unit 17A may, for example, generate summaries of the plurality of documents searched by the search unit 16A and select documents based on the generated summaries. In this case, the search result output unit 17A may, for example, select from the plurality of documents searched by the search unit 16A a document whose issue has a higher similarity to the issue extracted by the search unit 16A than other documents.

[0063] (Flow of diagnostic support method) Figure 8 is a flow diagram showing an example of the flow of a diagnostic support method S1A executed by information processing device 1A. The steps included in the flow diagram of Figure 8 may be executed in parallel or in a different order. In step S101, lesion detection unit 11A detects an abnormal area from diagnostic target image 202 of a patient who is the subject of diagnosis. In step S102, caption generation unit 12A generates a caption using information indicating the abnormal area detected by lesion detection unit 11A.

[0064] In step S103, the classification unit 13B generates a plurality of pieces of input information using the captions generated by the caption generation unit 12 A. In step S104, the discussion unit 14A generates a classification result for each of the generated pieces of input information.

[0065] In step S105, the discussion unit 14A outputs a discussion result from the standpoint of the multiple discrimination results. In step S106, the presentation unit 15A presents information indicating the differences between the multiple pieces of input information. In step S107, the search unit 16A searches for literature related to the discussion result. In step S108, the search result output unit 17A outputs the search results by the search unit 16A.

[0066] (Effects of the Information Processing Device) As described above, the information processing device 1A employs a configuration in which the multiple pieces of input information (first input information, second input information) input to the discrimination model are different from each other. Different input information can be used to vary the discrimination results, and discussions can be conducted based on the varied discrimination results. As a result, the information processing device 1A can obtain discussion results that consider the patient's possible medical condition from multiple angles based on the patient's medical information and lesion detection results, thereby more appropriately supporting the patient's diagnosis.

[0067] Furthermore, the information processing device 1A employs a configuration in which the input information includes a prompt including at least one of a caption, which is a sentence describing the abnormality, and medical information. Therefore, the information processing device 1A has an advantage in that, by using input information including at least one of a caption, which is a sentence describing the abnormality, and medical information, it is possible to obtain a more appropriate diagnosis result regarding the patient's condition.

[0068] Furthermore, the information processing device 1A employs a configuration in which the second input information includes some of the information included in the first input information, but does not include other information included in the first input information. By using multiple pieces of input information that share some of the same information, it is possible to vary the results of the discrimination, and discussions can be conducted based on these results. As a result, the information processing device 1A can obtain discussion results that are multifaceted and consider the patient's medical information, etc., and can more appropriately support the patient's diagnosis.

[0069] Furthermore, the information processing device 1A is configured to include a lesion detection unit 11A that detects abnormal areas from the diagnostic target image 202, and a caption generation unit 12A that generates captions indicating the abnormal areas detected by the lesion detection unit 11A. Therefore, the information processing device 1A can more appropriately differentiate the patient's condition by using input information including captions generated from the diagnostic target image 202 of the patient who is the diagnostic target.

[0070] Furthermore, the information processing device 1A employs a configuration in which the diagnostic object image 202 includes at least one of an X-ray image, an endoscopic image, a pathological image, an MRI image, and a CT image. Therefore, the information processing device 1A can obtain an effect of obtaining a more appropriate discrimination result by performing discrimination using at least one of an X-ray image, an endoscopic image, a pathological image, an MRI image, and a CT image.

[0071] Furthermore, the information processing device 1A employs a configuration in which the prompts included in the second input information include a portion of the prompts included in the first input information, but do not include other portions of the prompts included in the first input information. By using multiple pieces of input information that share a portion of the prompt, it is possible to vary the assessment results. This allows the information processing device 1A to output discussion results that have been considered from multiple angles based on multiple assessment results.

[0072] Furthermore, the information processing device 1A employs a configuration in which the first input information includes the diagnostic target image 202, and the second input information does not include the diagnostic target image. Therefore, the information processing device 1A can vary the discrimination results. This allows the information processing device 1A to output a discussion result that has been considered from multiple angles based on multiple appraisal results.

[0073] Furthermore, the information processing device 1A is configured such that the discrimination unit 13A outputs a third discrimination result regarding the patient's condition based on third input information including at least one of information indicating an abnormality detected from the diagnostic target image 202 and the medical information 201, and the discussion unit 14A outputs a discussion result based on the first discrimination result, the second discrimination result, and the third discrimination result. As a result, the information processing device 1A can output a discussion result that has been considered from multiple angles based on three or more multiple appraisal results.

[0074] The information processing device 1A also has a configuration including a presentation unit that presents information indicating the difference between the first input information and the second input information. By checking the presented information, a medical professional or the like diagnosing a patient can understand which difference in information caused the difference in the discrimination result. This allows the medical professional or the like to more appropriately diagnose the patient.

[0075] Furthermore, the information processing device 1A employs a configuration in which the discussion results output by the discussion unit 14A are information used to support decision-making regarding the diagnosis of a patient's condition, thereby enabling medical professionals and others diagnosing patients to make more appropriate decisions regarding the diagnosis.

[0076] The information processing device 1A also employs a configuration including a search unit 16A that extracts multiple points of discussion from the discussion results output by the discussion unit 14A and searches a database for literature related to the extracted points of discussion, and a search result output unit 17A that outputs the search results of the search unit 16A. By searching for literature closely related to the points of discussion identified by the discussion and outputting the search results, medical professionals and others diagnosing the patient can grasp literature related to the patient's condition, thereby enabling a more appropriate diagnosis of the patient.

[0077] In addition, in the information processing device 1A, the search result output unit 17A is configured to select one or more documents from the multiple documents searched by the search unit 16A and output the selected papers, thereby making it possible to select and output documents that are more closely related to the points of discussion identified through the discussion.

[0078] [Third Exemplary Embodiment] A third exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. Components having the same functions as those described in the above exemplary embodiment will be denoted by the same reference numerals, and their description will be omitted as appropriate. The scope of application of each technical means employed in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technical means employed in this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs. Furthermore, each technical means shown in each drawing referenced to describe this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs.

[0079] 9 is a block diagram showing the configuration of an information processing device 1B. The information processing device 1B includes a control unit 10B, a storage unit 20A, a communication unit 30A, an input unit 40A, and an output unit 50A. The control unit 10B includes a caption generation unit 12A, a discussion unit 14A, a presentation unit 15A, a search unit 16A, and a search result output unit 17A, as well as a lesion detection unit 11B and a differentiation unit 13B.

[0080] The lesion detection unit 11B acquires a diagnostic target image 202 of a patient who is the subject of diagnosis, and detects abnormal areas from the diagnostic target image 202. At this time, the lesion detection unit 11B detects abnormal areas using a plurality of different detection methods. Specifically, as an example, the lesion detection unit 11B inputs the diagnostic target image 202 into each of a plurality of detection models for detecting lesions, thereby acquiring a plurality of detection results.

[0081] The caption generating unit 12A generates a caption for each of the plurality of detection results by the lesion detecting unit 11B. The caption generating unit 12A supplies the generated plurality of captions to the distinguishing unit 13C.

[0082] The classification unit 13B generates input information including each of the multiple captions. In other words, of the multiple pieces of input information that the classification unit 13B inputs to the classification model, the first input information includes information indicating an abnormal location detected from the diagnostic target image by a first detection method, and the second input information includes information indicating an abnormal location detected from the diagnostic target image by a second detection method different from the first detection method. The classification unit 13B inputs the generated multiple pieces of input information to the classification model.

[0083] As described above, the information processing device 1B employs a configuration in which the first input information includes information indicating abnormal locations detected from the diagnostic target image using a first detection method, and the second input information includes information indicating abnormal locations detected from the diagnostic target image using a second detection method different from the first detection method. Therefore, the information processing device 1B can vary the discrimination results by using input information including information indicating abnormal locations detected using multiple different detection methods. By conducting a discussion using such multiple discrimination results, the information processing device 1B can obtain a discussion result that is considered from multiple angles based on multiple different detection results, thereby achieving the effect of more appropriately supporting the patient's diagnosis.

[0084] [Fourth Exemplary Embodiment] A fourth exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. Components having the same functions as those described in the above exemplary embodiment will be denoted by the same reference numerals, and their description will be omitted as appropriate. The scope of application of each technical means employed in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technical means employed in this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs. Furthermore, each technical means shown in each drawing referenced to describe this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs.

[0085] 10 is a block diagram showing the configuration of an information processing device 1C. The information processing device 1C includes a control unit 10C, a storage unit 20A, a communication unit 30A, an input unit 40A, and an output unit 50A. The control unit 10C includes a lesion detection unit 11A, a discussion unit 14A, a presentation unit 15A, a search unit 16A, and a search result output unit 17A, as well as a caption generation unit 12C and a classification unit 13C.

[0086] The caption generation unit 12C generates captions indicating abnormal areas detected by the lesion detection unit 11A. At this time, the caption generation unit 12C generates captions using a plurality of different generation methods. Specifically, as an example, the caption generation unit 12C generates a plurality of captions using a plurality of caption generation models for generating captions.

[0087] 11 is a diagram showing a specific example of multiple captions generated by caption generation unit 12C. In the example of Fig. 11, caption generation unit 12C inputs image 203, on which lesion detection locations are marked and output from lesion detection model M1, to caption generation models M2a and M2b. Caption generation model M2a outputs caption 204a, while caption generation model M2b outputs caption 204b.

[0088] The classification unit 13C generates input information including each of the plurality of captions. In other words, among the plurality of pieces of input information input to the classification model by the classification unit 13C, the first input information includes a caption generated from the information indicating the abnormality location by a first generation method, and the second input information includes a caption generated from the information indicating the abnormality location by a second generation method different from the first generation method.

[0089] As described above, the information processing device 1C employs a configuration in which the first input information includes a caption generated from information indicating an abnormality using a first generation method, and the second input information includes a caption generated from information indicating an abnormality using a second generation method different from the first generation method. Therefore, the information processing device 1C can vary the discrimination results by using input information including captions generated using multiple different generation methods. By conducting a discussion using such multiple discrimination results, the information processing device 1C can obtain a discussion result that is considered from multiple angles based on multiple different captions, thereby achieving the effect of more appropriately supporting patient diagnosis.

[0090] Fifth Exemplary Embodiment A fifth exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. Components having the same functions as those described in the above exemplary embodiments will be denoted by the same reference numerals, and their description will be omitted as appropriate. The scope of application of each technical means employed in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technical means employed in this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs. Furthermore, each technical means shown in each drawing referenced to describe this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs.

[0091] 12 is a block diagram showing the configuration of an information processing device 1D. The information processing device 1D includes a control unit 10D, a storage unit 20A, a communication unit 30A, an input unit 40A, and an output unit 50A. The control unit 10D includes a lesion detection unit 11A, a caption generation unit 12A, a discussion unit 14A, a presentation unit 15A, a search unit 16A, and a search result output unit 17A, as well as a discrimination unit 13D.

[0092] The discrimination unit 13D outputs a discrimination result regarding the patient's condition. At this time, the discrimination unit 13D outputs the discrimination results discriminated by a plurality of different discrimination methods. Specific examples of the discrimination methods are described below. The discrimination unit 13D may use a combination of the discrimination methods described below.

[0093] (Example 1 of Classification Method) As an example, the classification unit 13D inputs input information to multiple classification models to obtain multiple classification results. In other words, the classification unit 13D outputs a first classification result obtained by inputting the first input information to a first trained model (classification model) generated by machine learning, and also outputs a second classification result obtained by inputting the second input information to a second trained model (classification model) generated by machine learning and different from the first trained model. In this case, the second input information may be the same as the first input information or may be different from the first input information.

[0094] 13 is a diagram showing a specific example of a classification result output by the classification unit 13D. In the example of FIG. 13, the classification unit 13D inputs a prompt 215 to classification models M3a and M3b. The classification model M3a outputs a classification result 216a. Meanwhile, the classification model M3b outputs a classification result 216b.

[0095] (Example 2 of discrimination method) Figure 14 is a diagram showing another example of a discrimination method in the discrimination unit 13D. In the example of Figure 14, the discrimination unit 13D inputs a feature vector 218 obtained by inputting a diagnostic target image 202 into an image encoder 217, and a feature vector 220 obtained by inputting a prompt 215 including medical information 201 into a text encoder 219, into a discrimination model M3c. The discrimination model M3c outputs a diagnosis result 216c. The image encoder 217 generates a feature vector representing the image features from the image data. The text encoder 219 generates a feature vector representing the text features from the text data. The image encoder 217 and the text encoder 219 are, for example, neural networks generated by machine learning, but are not limited to this.

[0096] (Example 3 of Discrimination Method) Figure 15 is a diagram showing another example of the discrimination method in the discrimination unit 13D. In the example of Figure 15, the discrimination unit 13D inputs a feature vector 221 obtained by inputting an image 203 with a marked lesion site into an image encoder 217, and a feature vector 220 obtained by inputting a prompt 215 including medical information 201 into a text encoder 219, to a discrimination model M3c. The discrimination model M3c outputs a diagnosis result 216d.

[0097] As described above, the information processing device 1D is configured to output a first discrimination result obtained by inputting first input information into a first discrimination model generated by machine learning, and to output a second discrimination result obtained by inputting second input information into a second discrimination model generated by machine learning and different from the first discrimination model. Therefore, the information processing device 1D can vary the discrimination results by using multiple different discrimination models. By conducting a discussion using such multiple discrimination results, the information processing device 1D can obtain a discussion result that is considered from multiple angles based on the multiple different discrimination results, thereby achieving the effect of more appropriately supporting patient diagnosis.

[0098] [Example of implementation by software] Some or all of the functions of the diagnostic support device 1 and the information processing devices 1A, 1B, 1C, and 1D (hereinafter also referred to as "each of the above-mentioned devices") may be implemented by hardware such as an integrated circuit (IC chip), or by software.

[0099] In the latter case, each of the above devices is realized by, for example, a computer that executes instructions of a program, which is software that realizes each function. An example of such a computer (hereinafter referred to as computer C) is shown in Figure 16. Figure 16 is a block diagram showing the hardware configuration of computer C that functions as each of the above devices.

[0100] The computer C includes at least one processor C1 and at least one memory C2. The memory C2 stores a program P for causing the computer C to function as each of the above-mentioned devices. In the computer C, the processor C1 reads and executes the program P from the memory C2, thereby realizing the functions of each of the above-mentioned devices.

[0101] The processor C1 may be, for example, a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), a micro processing unit (MPU), a floating point number processing unit (FPU), a physics processing unit (PPU), a tensor processing unit (TPU), a quantum processor, a microcontroller, or a combination thereof. The memory C2 may be, for example, a flash memory, a hard disk drive (HDD), a solid state drive (SSD), or a combination thereof.

[0102] The computer C may further include a RAM (Random Access Memory) for expanding the program P during execution and for temporarily storing various data. The computer C may also include a communication interface for transmitting and receiving data to and from other devices. The computer C may also include an input / output interface for connecting input / output devices such as a keyboard, a mouse, a display, and a printer.

[0103] The program P can also be recorded on a non-transitory, tangible recording medium M that can be read by the computer C. Such a recording medium M can be, for example, a tape, a disk, a card, a semiconductor memory, or a programmable logic circuit. The computer C can acquire the program P via such a recording medium M. The program P can also be transmitted via a transmission medium. Such a transmission medium can be, for example, a communication network or broadcast waves. The computer C can also acquire the program P via such a transmission medium.

[0104] [Appendix 1] This disclosure includes the techniques described in the following appendices. However, the present invention is not limited to the techniques described in the following appendices, and various modifications are possible within the scope of the claims.

[0105] [Appendix A] This disclosure includes the techniques described in the following appendices. However, the present invention is not limited to the techniques described in the following appendices, and various modifications are possible within the scope of the claims.

[0106] (Appendix A1) A diagnostic support device comprising: a first discrimination means for outputting a first discrimination result regarding the condition of a patient to be diagnosed based on first input information including at least one of information indicating an abnormal location detected from a diagnostic target image of the patient and medical information of the patient; a second discrimination means for outputting a second discrimination result regarding the condition of the patient based on second input information including at least one of information indicating an abnormal location detected from the diagnostic target image and the medical information; and a discussion means for outputting a discussion result indicating the content of a discussion from the standpoint of the first discrimination result and the standpoint of the second discrimination result when the first discrimination result and the second discrimination result differ.

[0107] (Appendix A2) The diagnosis support device according to Appendix A1, wherein the second input information is information different from the first input information.

[0108] (Appendix A3) The diagnosis support device according to Appendix A1 or A2, wherein the first input information and the second input information each include a prompt including at least one of a caption that is a sentence explaining the abnormality location and the medical information.

[0109] (Appendix A4) The diagnosis support device according to Appendix A2, wherein the second input information includes a portion of information included in the first input information, and does not include another portion of information included in the first input information.

[0110] (Appendix A5) The diagnostic support device described in any one of Appendices A2 to A4, wherein the first input information includes information indicating an abnormal area detected from the diagnostic target image by a first detection method, and the second input information includes information indicating an abnormal area detected from the diagnostic target image by a second detection method different from the first detection method.

[0111] (Appendix A6) The diagnostic support device described in any one of Appendices A2 to A5, wherein the first input information includes a caption generated from information indicating an abnormality location by a first generation method, and the second input information includes a caption generated from information indicating an abnormality location by a second generation method different from the first generation method.

[0112] (Appendix A7) The diagnostic support device described in any one of Appendices A1 to A6, wherein the first discrimination means outputs the first discrimination result obtained by inputting the first input information into a first trained model generated by machine learning, and the second discrimination means outputs a second discrimination result obtained by inputting the second input information into a second trained model generated by machine learning, the second trained model being different from the first trained model.

[0113] (Appendix A8) The diagnosis support device according to Appendix A3, further comprising: a detection means for detecting the abnormal portion from the diagnostic target image; and a caption generation means for generating the caption indicating the abnormal portion detected by the detection means.

[0114] (Supplementary Note A9) The diagnosis support device according to any one of Supplementary Notes A1 to A8, wherein the diagnostic object image includes at least one of an X-ray image, an endoscopic image, a pathological image, an MRI image, and a CT image.

[0115] (Appendix A10) The diagnostic support device according to appendix A3 or A8, wherein the prompt included in the second input information includes a part of the prompt included in the first input information, and does not include information on another part of the prompt included in the first input information.

[0116] (Supplementary Note A11) The diagnosis support device according to any one of Supplementary Notes A1 to A10, wherein the first input information includes the diagnostic object image, and the second input information does not include the diagnostic object image.

[0117] (Appendix A12) The diagnostic support device described in any one of Appendices A1 to A11, wherein the second discrimination means outputs a third discrimination result regarding the patient's condition based on third input information including at least one of information indicating abnormal areas detected from the diagnostic target image and the medical information, and the discussion means outputs a result of a discussion based on the first discrimination result, the second discrimination result, and the third discrimination result.

[0118] (Supplementary Note A13) The diagnosis support device according to Supplementary Note A2, further comprising: a presentation unit that presents information indicating a difference between the first input information and the second input information.

[0119] (Supplementary Note A14) The diagnosis support device according to any one of Supplementary Notes A1 to A13, wherein the discussion result is information used to support decision-making regarding a diagnosis of the patient's condition.

[0120] (Appendix A15) A diagnostic support device according to any one of Appendices A1 to A14, further comprising: a search means for extracting a plurality of points of discussion from the discussion results and searching a database for literature related to the extracted points of discussion; and a search result output means for outputting the search results obtained by the search means.

[0121] (Appendix A16) The diagnosis support device according to Appendix A15, wherein the search result output means selects one or more documents from the plurality of documents searched by the search means, and outputs the selected documents.

[0122] [Appendix B] This disclosure includes the techniques described in the following appendices. However, the present invention is not limited to the techniques described in the following appendices, and various modifications are possible within the scope of the claims.

[0123] (Appendix B1) A diagnostic support method comprising: a first discrimination process in which at least one processor outputs a first discrimination result regarding the patient's condition based on first input information including at least one of information indicating an abnormal location detected from a diagnostic target image of the patient who is the subject of diagnosis and medical information of the patient; a second discrimination process in which the at least one processor outputs a second discrimination result regarding the patient's condition based on second input information including at least one of information indicating an abnormal location detected from the diagnostic target image and the medical information; and a discussion process in which the at least one processor outputs a discussion result indicating the content of a discussion from the standpoint of the first discrimination result and the standpoint of the second discrimination result when the first discrimination result and the second discrimination result differ.

[0124] (Supplementary Note B2) The diagnostic support method according to Supplementary Note B1, wherein the second input information is information different from the first input information.

[0125] (Appendix B3) The diagnostic support method according to Appendix B1 or B2, wherein the first input information and the second input information each include a prompt including at least one of a caption that is a sentence explaining the abnormality location and the medical information.

[0126] (Supplementary Note B4) The diagnostic support method according to Supplementary Note B2, wherein the second input information includes some of the information included in the first input information, and does not include other part of the information included in the first input information.

[0127] (Appendix B5) A diagnostic support method described in any one of Appendices B2 to B4, wherein the first input information includes information indicating an abnormal area detected from the diagnostic target image by a first detection method, and the second input information includes information indicating an abnormal area detected from the diagnostic target image by a second detection method different from the first detection method.

[0128] (Appendix B6) A diagnostic support method described in any one of Appendices B2 to B5, wherein the first input information includes a caption generated from information indicating an abnormality location by a first generation method, and the second input information includes a caption generated from information indicating an abnormality location by a second generation method different from the first generation method.

[0129] (Appendix B7) A diagnostic support method described in any one of Appendices B1 to B6, wherein in the first discrimination process, the at least one processor outputs the first discrimination result obtained by inputting the first input information into a first trained model generated by machine learning, and in the second discrimination process, the at least one processor outputs a second discrimination result obtained by inputting the second input information into a second trained model generated by machine learning, the second trained model being different from the first trained model.

[0130] (Appendix B8) The diagnostic support method described in Appendix B3 further includes: a detection process in which the at least one processor detects the abnormal area from the diagnostic target image; and a caption generation process in which the at least one processor generates the caption indicating the abnormal area detected in the detection process.

[0131] (Supplementary Note B9) The diagnostic support method according to any one of Supplementary Notes B1 to B8, wherein the diagnostic target image includes at least one of an X-ray image, an endoscopic image, a pathological image, an MRI image, and a CT image.

[0132] (Appendix B10) The diagnostic support method described in Appendix B3 or B8, wherein the prompt included in the second input information includes a part of the prompt included in the first input information, and does not include information on another part of the prompt included in the first input information.

[0133] (Supplementary Note B11) The diagnostic support method according to any one of Supplementary Notes B1 to B10, wherein the first input information includes the diagnostic object image, and the second input information does not include the diagnostic object image.

[0134] (Appendix B12) A diagnostic support method described in any one of Appendices B1 to B11, wherein in the second discrimination process, the at least one processor outputs a third discrimination result regarding the patient's condition based on third input information including at least one of information indicating abnormal areas detected from the diagnostic target image and the medical information, and in the discussion process, the at least one processor outputs a result of a discussion based on the first discrimination result, the second discrimination result, and the third discrimination result.

[0135] (Supplementary Note B13) The diagnostic support method according to Supplementary Note B2, further comprising a presentation process in which the at least one processor presents information indicating a difference between the first input information and the second input information.

[0136] (Supplementary Note B14) The diagnostic support method according to any one of Supplementary Notes B1 to B13, wherein the discussion result is information used to support decision-making regarding a diagnosis of the patient's condition.

[0137] (Appendix B15) A diagnostic support method described in any one of Appendices B1 to B14, further comprising: a search process in which the at least one processor extracts multiple points of discussion from the discussion results and searches a database for literature related to the extracted points of discussion; and a search result output process in which the at least one processor outputs search results from the search process.

[0138] (Appendix B16) The diagnostic support method according to Appendix B15, wherein in the search result output process, the at least one processor selects one or more documents from the plurality of documents searched in the search process, and outputs the selected documents.

[0139] [Appendix C] This disclosure includes the techniques described in the following appendices. However, the present invention is not limited to the techniques described in the following appendices, and various modifications are possible within the scope of the claims.

[0140] (Appendix C1) A diagnostic support program for causing a computer to function as a diagnostic support device, the diagnostic support program causing the computer to function as: a first discrimination means for outputting a first discrimination result regarding the condition of a patient based on first input information including at least one of information indicating an abnormality detected from a diagnostic target image of the patient and medical information of the patient; a second discrimination means for outputting a second discrimination result regarding the condition of the patient based on second input information including at least one of information indicating an abnormality detected from the diagnostic target image and the medical information; and a discussion means for outputting a discussion result indicating the content of a discussion from the standpoint of the first discrimination result and the standpoint of the second discrimination result when the first discrimination result and the second discrimination result differ.

[0141] (Supplementary Note C2) The diagnostic assistance program according to Supplementary Note C1, wherein the second input information is information different from the first input information.

[0142] (Appendix C3) The diagnostic assistance program according to Appendix C1 or C2, wherein the first input information and the second input information each include a prompt including at least one of a caption that is a sentence explaining the abnormality and the medical information.

[0143] (Supplementary Note C4) The diagnostic assistance program according to Supplementary Note C2, wherein the second input information includes some of the information included in the first input information, and does not include other part of the information included in the first input information.

[0144] (Appendix C5) A diagnostic assistance program described in any one of Appendices C2 to C4, wherein the first input information includes information indicating an abnormal area detected from the diagnostic target image by a first detection method, and the second input information includes information indicating an abnormal area detected from the diagnostic target image by a second detection method different from the first detection method.

[0145] (Appendix C6) A diagnostic assistance program described in any one of Appendices C2 to C5, wherein the first input information includes a caption generated from information indicating an abnormality location using a first generation method, and the second input information includes a caption generated from information indicating an abnormality location using a second generation method different from the first generation method.

[0146] (Appendix C7) A diagnostic assistance program described in any one of Appendices C1 to C6, wherein the first discrimination means outputs the first discrimination result obtained by inputting the first input information into a first trained model generated by machine learning, and the second discrimination means outputs a second discrimination result obtained by inputting the second input information into a second trained model generated by machine learning and different from the first trained model.

[0147] (Appendix C8) A diagnostic assistance program described in Appendix C3, which further causes the computer to function as a detection means for detecting the abnormal area from the diagnostic target image, and a caption generation means for generating the caption indicating the abnormal area detected by the detection means.

[0148] (Supplementary Note C9) The diagnostic support program according to any one of Supplementary Notes C1 to C8, wherein the diagnostic target image includes at least one of an X-ray image, an endoscopic image, a pathological image, an MRI image, and a CT image.

[0149] (Appendix C10) A diagnostic assistance program as described in Appendix C3 or C8, wherein the prompt included in the second input information includes a part of the prompt included in the first input information, and does not include information on another part of the prompt included in the first input information.

[0150] (Supplementary Note C11) The diagnostic assistance program according to any one of Supplementary Notes C1 to C10, wherein the first input information includes the diagnostic target image, and the second input information does not include the diagnostic target image.

[0151] (Appendix C12) A diagnostic assistance program described in any one of Appendices C1 to C11, wherein the second discrimination means outputs a third discrimination result regarding the patient's condition based on third input information including at least one of information indicating abnormal areas detected from the diagnostic target image and the medical information, and the discussion means outputs a result of a discussion based on the first discrimination result, the second discrimination result, and the third discrimination result.

[0152] (Appendix C13) The diagnostic assistance program according to Appendix C2, further causing the computer to function as a presentation unit that presents information indicating a difference between the first input information and the second input information.

[0153] (Appendix C14) The diagnostic support program according to any one of appendices C1 to C13, wherein the discussion result is information used to support decision-making regarding a diagnosis of the patient's condition.

[0154] (Appendix C15) A diagnostic support program described in any one of Appendices C1 to C14, further causing the computer to function as a search means for extracting multiple points of discussion from the discussion results and searching a database for literature related to the extracted points of discussion, and a search result output means for outputting search results obtained by the search means.

[0155] (Appendix C16) The diagnostic support program according to Appendix C15, wherein the search result output means selects one or more documents from the plurality of documents searched by the search means, and outputs the selected documents.

[0156] [Appendix D] This disclosure includes the techniques described in the following appendices. However, the present invention is not limited to the techniques described in the following appendices, and various modifications are possible within the scope of the claims.

[0157] (Appendix D1) A diagnostic support device comprising at least one processor, the at least one processor executing: a first discrimination process that outputs a first discrimination result regarding the condition of a patient to be diagnosed based on first input information including at least one of information indicating an abnormal location detected from a diagnostic target image of the patient and medical information of the patient; a second discrimination process that outputs a second discrimination result regarding the condition of the patient based on second input information including at least one of information indicating an abnormal location detected from the diagnostic target image and the medical information; and a discussion process that outputs a discussion result indicating the content of a discussion from the standpoint of the first discrimination result and the standpoint of the second discrimination result when the first discrimination result and the second discrimination result differ.

[0158] The diagnosis support device may further include a memory, and the memory may store a program for causing the at least one processor to execute each of the processes.

[0159] (Appendix D2) The diagnosis support device according to appendix D1, wherein the second input information is information different from the first input information.

[0160] (Appendix D3) The diagnosis support device according to Appendix D1 or D2, wherein the first input information and the second input information each include a prompt including at least one of a caption that is a sentence explaining the abnormality and the medical information.

[0161] (Appendix D4) The diagnosis support device according to Appendix D2, wherein the second input information includes a portion of information included in the first input information, and does not include another portion of information included in the first input information.

[0162] (Appendix D5) A diagnostic support device described in any one of Appendices D2 to D4, wherein the first input information includes information indicating an abnormal area detected from the diagnostic target image by a first detection method, and the second input information includes information indicating an abnormal area detected from the diagnostic target image by a second detection method different from the first detection method.

[0163] (Appendix D6) A diagnostic support device described in any one of Appendices D2 to D5, wherein the first input information includes a caption generated from information indicating an abnormality location by a first generation method, and the second input information includes a caption generated from information indicating an abnormality location by a second generation method different from the first generation method.

[0164] (Appendix D7) A diagnostic support device described in any one of Appendices D1 to D6, wherein in the first discrimination process, the at least one processor outputs the first discrimination result obtained by inputting the first input information into a first trained model generated by machine learning, and in the second discrimination process, the at least one processor outputs a second discrimination result obtained by inputting the second input information into a second trained model generated by machine learning, the second trained model being different from the first trained model.

[0165] (Appendix D8) The diagnostic support device described in Appendix D3 further executes a detection process for detecting the abnormal area from the diagnostic target image, and a caption generation process in which the at least one processor generates a caption indicating the abnormal area detected in the detection process.

[0166] (Appendix D9) The diagnosis support device according to any one of appendices D1 to D8, wherein the diagnostic target image includes at least one of an X-ray image, an endoscopic image, a pathological image, an MRI image, and a CT image.

[0167] (Appendix D10) The diagnostic support device described in Appendix D3 or D8, wherein the prompt included in the second input information includes a part of the prompt included in the first input information, and does not include information on another part of the prompt included in the first input information.

[0168] (Appendix D11) The diagnosis support device according to any one of appendices D1 to D10, wherein the first input information includes the diagnostic target image, and the second input information does not include the diagnostic target image.

[0169] (Appendix D12) In the second discrimination process, the at least one processor outputs a third discrimination result regarding the patient's condition based on third input information including at least one of information indicating abnormal areas detected from the diagnostic target image and the medical information; and in the discussion process, the at least one processor outputs a result of a discussion based on the first discrimination result, the second discrimination result, and the third discrimination result. A diagnostic support device described in any one of Appendices D1 to D11.

[0170] (Supplementary Note D13) The diagnosis support device according to Supplementary Note D2, wherein the at least one processor further executes a presentation process of presenting information indicating a difference between the first input information and the second input information.

[0171] (Supplementary Note D14) The diagnosis support device according to any one of Supplementary Notes D1 to D13, wherein the discussion result is information used to support decision-making regarding a diagnosis of the patient's condition.

[0172] (Appendix D15) A diagnostic support device described in any one of Appendices D1 to D14, wherein the at least one processor further executes a search process that extracts multiple points of discussion from the discussion results and searches a database for literature related to the extracted points of discussion, and a search result output process that outputs the search results obtained by the search process.

[0173] (Appendix D16) The diagnostic support device according to Appendix D15, wherein in the search result output process, the at least one processor selects one or more documents from the plurality of documents searched in the search process, and outputs the selected documents.

[0174] [Appendix E] This disclosure includes the techniques described in the following appendices. However, the present invention is not limited to the techniques described in the following appendices, and various modifications are possible within the scope of the claims.

[0175] (Appendix E1) A non-transient recording medium having recorded thereon a diagnostic assistance program for causing a computer to function as a diagnostic assistance device, the diagnostic assistance program causing the computer to execute the following: a first discrimination process for outputting a first discrimination result regarding the patient's condition based on first input information including at least one of information indicating an abnormality detected from a diagnostic target image of the patient who is the subject of diagnosis and medical information of the patient; a second discrimination process for outputting a second discrimination result regarding the patient's condition based on second input information including at least one of information indicating an abnormality detected from the diagnostic target image and the medical information; and a discussion process for outputting a discussion result indicating the content of a discussion from the standpoint of the first discrimination result and the standpoint of the second discrimination result when the first discrimination result and the second discrimination result differ.

[0176] REFERENCE SIGNS LIST 1 Diagnosis support device 1A, 1B, 1C, 1D Information processing device 11 First identification unit 11A, 11B Lesion detection unit 12 Second identification unit 12A, 12C Caption generation unit 13, 14A Discussion unit 13A, 13B, 13C, 13D Identification unit 15A Presentation unit 16A Search unit 17A Search result output unit

Claims

1. A diagnostic support device comprising: a first discrimination means for outputting a first discrimination result regarding the medical condition of the patient based on first input information including at least one of information indicating an abnormal location detected from a diagnostic target image of the patient to be diagnosed and the medical information of the patient; a second discrimination means for outputting a second discrimination result regarding the medical condition of the patient based on second input information including at least one of information indicating an abnormal location detected from the diagnostic target image and the medical information; and a discussion means for outputting a discussion result indicating the content of the discussion from the standpoint of the first discrimination result and the standpoint of the second discrimination result when the first discrimination result and the second discrimination result are different.

2. The diagnostic support device according to claim 1, wherein the second input information is information different from the first input information.

3. The diagnostic support device according to claim 1 or 2, wherein the first input information and the second input information each include a caption that is a sentence explaining the abnormal location and a prompt including at least one of the medical information.

4. The diagnostic support device according to claim 2, wherein the second input information includes some of the information included in the first input information and does not include some other information included in the first input information.

5. The diagnostic support device according to any one of claims 2 to 4, wherein the first input information includes information indicating an abnormal location detected from the diagnostic target image by a first detection method, and the second input information includes information indicating an abnormal location detected from the diagnostic target image by a second detection method different from the first detection method.

6. The diagnostic support device according to any one of claims 2 to 5, wherein the first input information includes a caption generated by a first generation method from information indicating an abnormal location, and the second input information includes a caption generated by a second generation method different from the first generation method from information indicating an abnormal location.

7. The first discrimination means outputs the first discrimination result obtained by inputting the first input information into a first learned model generated by machine learning, and the second discrimination means is a second learned model generated by machine learning and different from the first learned model, and outputs a second discrimination result obtained by inputting the second input information into the second learned model. The diagnostic support device according to any one of claims 1 to 6.

8. The diagnostic support device according to claim 3, further comprising: detection means for detecting the abnormal portion from the diagnostic target image; and caption generation means for generating a caption indicating the abnormal portion detected by the detection means.

9. The diagnostic support device according to any one of claims 1 to 8, wherein the diagnostic target image includes at least one of an X-ray image, an endoscopic image, a pathological image, an MRI image, and a CT image.

10. The diagnostic support device according to claim 3 or 8, wherein the prompt included in the second input information includes a part of the prompt included in the first input information and does not include other part information of the prompt included in the first input information.

11. The diagnostic support device according to any one of claims 1 to 10, wherein the first input information includes the diagnostic target image, and the second input information does not include the diagnostic target image.

12. The second discrimination means outputs a third discrimination result regarding the patient's medical condition based on third input information including at least one of information indicating an abnormal portion detected from the diagnostic target image and the medical information, and the discussion means outputs a result of discussion based on the first discrimination result, the second discrimination result, and the third discrimination result. The diagnostic support device according to any one of claims 1 to 11.

13. The diagnostic support device according to claim 2, further comprising presentation means for presenting information indicating a difference between the first input information and the second input information.

14. The diagnostic support device according to any one of claims 1 to 13, wherein the discussion result is information used to support a decision-making regarding the diagnosis of the patient's medical condition.

15. A diagnostic support apparatus according to any one of claims 1 to 14, further comprising: a search means for extracting a plurality of arguments from the discussion results and searching a database for documents related to the extracted arguments; and a search result output means for outputting the search results obtained by the search means.

16. The diagnostic support apparatus according to claim 15, wherein the search result output means selects one or more documents from among the plurality of documents searched by the search means and outputs the selected documents.

17. A diagnostic support method including: a first discrimination process in which at least one processor outputs a first discrimination result regarding the medical condition of a patient based on first input information including at least one of information indicating an abnormal portion detected from a diagnostic target image of the patient to be diagnosed and the medical information of the patient; a second discrimination process in which the at least one processor outputs a second discrimination result regarding the medical condition of the patient based on second input information including at least one of information indicating an abnormal portion detected from the diagnostic target image and the medical information; and a discussion process in which the at least one processor outputs a discussion result indicating the content of the discussion from the standpoint of the first discrimination result and the standpoint of the second discrimination result when the first discrimination result and the second discrimination result are different.

18. A program for causing a computer to function as a diagnostic support program, the program causing the computer to function as: a first discrimination means for outputting a first discrimination result regarding the medical condition of a patient based on first input information including at least one of information indicating an abnormal portion detected from a diagnostic target image of the patient to be diagnosed and the medical information of the patient; a second discrimination means for outputting a second discrimination result regarding the medical condition of the patient based on second input information including at least one of information indicating an abnormal portion detected from the diagnostic target image and the medical information; and a discussion means for outputting a discussion result indicating the content of the discussion from the standpoint of the first discrimination result and the standpoint of the second discrimination result when the first discrimination result and the second discrimination result are different.

Images