Information processing system and information processing apparatus

Abstract

The present application provides an information processing system and an information processing device, the information processing system having an image acquisition device that acquires a patient's eye image data, and a first information processing device that can communicate with the image acquisition device and stores the eye image data, the image acquisition device transmitting first transmission data to the first information processing device, the first transmission data including the eye image data and additional information for determining an image diagnostic device that performs image diagnosis on the eye image data, the first information processing device, if it receives the first transmission data from the image acquisition device, storing the eye image data, and based on the additional information, determining at least one of a first image diagnostic device that performs a first image diagnosis on the eye image data and a second image diagnostic device that performs a second image diagnosis different from the first image diagnosis on the eye image data, and transmitting second transmission data including the eye image data to the determined image diagnostic device.

Description

[0001] This invention is a divisional application of the invention application filed on July 29, 2020, with international application number PCT / JP2020 / 029055, national application number 202080068885.2 which entered the Chinese national phase, and entitled "Information Processing System, Information Processing Apparatus, Information Processing Method and Program". Technical Field

[0002] This invention relates to information processing systems and information processing devices. Background Technology

[0003] An ophthalmic information processing server capable of performing ophthalmic image analysis is known (see Patent Document 1). However, in the past, the selection of an appropriate ophthalmic information processing server based on factors such as the device that captured the image or the client who commissioned the image analysis has not been considered.

[0004] Existing technical documents

[0005] Patent documents

[0006] Patent Document 1: Japanese Patent No. 5951086 Summary of the Invention

[0007] An information processing system comprising one technical solution of the invention disclosed in this application includes: an image acquisition device for acquiring image data of a patient's eye to be examined; and a first information processing device capable of communicating with the image acquisition device and storing the image data of the eye to be examined. The image acquisition device performs a first transmission process, which transmits first transmission data, including the image data of the eye to be examined and additional information for determining an image diagnostic device that performs an image diagnosis on the image data of the eye to be examined, to the first information processing device. The first information processing device performs the following processes: a storage process, which stores the image data of the eye to be examined if the first transmission data is received from the image acquisition device; a determination process, which determines at least one of a first image diagnostic device and a second image diagnostic device based on the additional information, wherein the first image diagnostic device performs a first image diagnosis on the image data of the eye to be examined, and the second image diagnostic device performs a second image diagnosis on the image data of the eye to be examined that is different from the first image diagnosis; and a second transmission process, which transmits second transmission data, including the image data of the eye to be examined, to the determined image diagnostic device.

[0008] An information processing system comprising one technical solution of the invention disclosed in this application includes: an image acquisition device for acquiring image data of a patient's examined eye; and a first information processing device capable of communicating with the image acquisition device and storing the examined eye image data. The image acquisition device performs a first transmission process, which transmits first transmission data, including the examined eye image data and additional information for determining artificial intelligence for image diagnosis of the examined eye image data, to the first information processing device. The first information processing device performs the following processes: a storage process, which stores the examined eye image data if the first transmission data is received from the image acquisition device; a determination process, which determines at least one of a first artificial intelligence and a second artificial intelligence based on the additional information, wherein the first artificial intelligence is used to perform a first image diagnosis of the examined eye image data, and the second artificial intelligence performs a second image diagnosis of the examined eye image data that is different from the first image diagnosis; and a second transmission process, which transmits second transmission data, including the examined eye image data and determination information for determining the artificial intelligence, to an image diagnosis device containing the determined artificial intelligence.

[0009] An information processing apparatus comprising one technical solution of the invention disclosed in this application includes a processor and a storage device. The storage device stores patient eye image data, additional information of the eye image data, and correspondence information between the additional information and an image diagnostic device that performs image diagnosis on the eye image data. The processor determines at least one of a first image diagnostic device and a second image diagnostic device based on the correspondence information and the additional information of the eye image data. The first image diagnostic device performs a first image diagnosis on the eye image data, and the second image diagnostic device performs a second image diagnosis on the eye image data that is different from the first image diagnosis. The processor sends transmission data containing the eye image data to the determined image diagnostic device.

[0010] An information processing apparatus comprising one technical solution of the invention disclosed in this application includes a processor and a storage device. The storage device stores patient eye image data, additional information of the eye image data, and correspondence information between the additional information and artificial intelligence performing image diagnosis on the eye image data. The processor determines at least one of a first artificial intelligence and a second artificial intelligence based on the correspondence information and the additional information of the eye image data. The first artificial intelligence is used to perform a first image diagnosis on the eye image data, and the second artificial intelligence performs a second image diagnosis on the eye image data that is different from the first image diagnosis. The processor sends second transmission data containing the eye image data and determination information of the determined artificial intelligence to an image diagnosis device containing the determined artificial intelligence.

[0011] The information processing method, which is a technical solution of the invention disclosed in this application, is executed by an information processing apparatus. The information processing apparatus includes a processor and a storage device. The storage device stores patient's examined eye image data, additional information of the examined eye image data, and correspondence information between the additional information and an image diagnostic device that performs image diagnosis on the examined eye image data. In the information processing method, the processor determines at least one of a first image diagnostic device and a second image diagnostic device based on the correspondence information and the additional information of the examined eye image data. The first image diagnostic device performs a first image diagnosis on the examined eye image data, and the second image diagnostic device performs a second image diagnosis on the examined eye image data that is different from the first image diagnosis. The processor sends transmission data containing the examined eye image data to the determined image diagnostic device.

[0012] The information processing method, which is a technical solution of the invention disclosed in this application, is executed by an information processing apparatus. The information processing apparatus includes a processor and a storage device. The storage device stores patient eye image data, additional information of the eye image data, and correspondence information between the additional information and artificial intelligence performing image diagnosis on the eye image data. In the information processing method, the processor determines at least one of a first artificial intelligence and a second artificial intelligence based on the correspondence information and the additional information of the eye image data. The first artificial intelligence is used to perform a first image diagnosis of the eye image data, and the second artificial intelligence performs a second image diagnosis of the eye image data that is different from the first image diagnosis. The processor sends second transmission data, which includes the eye image data and determination information of the determined artificial intelligence, to an image diagnosis device containing the determined artificial intelligence.

[0013] A procedure that constitutes a technical solution of the invention disclosed in this application is a procedure for causing an information processing device to perform information processing. The information processing device includes a processor and a storage device. The storage device stores patient-examined eye image data, additional information of the examined eye image data, and correspondence information between the additional information and an image diagnostic device that performs image diagnosis on the examined eye image data. The procedure causes the processor to perform the following processing: determining at least one of a first image diagnostic device and a second image diagnostic device based on the correspondence information and the additional information of the examined eye image data, wherein the first image diagnostic device performs a first image diagnosis on the examined eye image data, and the second image diagnostic device performs a second image diagnosis on the examined eye image data that is different from the first image diagnosis; and sending transmission data containing the examined eye image data to the determined image diagnostic device.

[0014] A procedure that constitutes a technical solution of the invention disclosed in this application is a procedure for causing an information processing device to perform information processing. The information processing device includes a processor and a storage device. The storage device stores patient eye image data, additional information of the eye image data, and correspondence information between the additional information and artificial intelligence performing image diagnosis on the eye image data. The procedure causes the processor to perform the following processing: determining at least one of a first artificial intelligence and a second artificial intelligence based on the correspondence information and the additional information of the eye image data, wherein the first artificial intelligence is used to perform a first image diagnosis of the eye image data, and the second artificial intelligence performs a second image diagnosis of the eye image data that is different from the first image diagnosis; and sending second transmission data containing the eye image data and determination information of the determined artificial intelligence to an image diagnosis device containing the determined artificial intelligence. Attached Figure Description

[0015] Figure 1 This is an explanatory diagram showing a structural example of the image diagnostic system in Embodiment 1.

[0016] Figure 2 This is a block diagram illustrating an example of the hardware structure of the computer in Embodiment 1.

[0017] Figure 3 This is a block diagram illustrating the functional structure of the management server in Embodiment 1.

[0018] Figure 4 This is a block diagram illustrating the functional structure of the diagnostic server in Embodiment 1.

[0019] Figure 5 This is a block diagram illustrating the functional structure of the in-hospital server in Embodiment 1.

[0020] Figure 6 This is a block diagram illustrating the functional structure of the terminal in Embodiment 1.

[0021] Figure 7 This is a timing diagram illustrating the image diagnostic processing of the image diagnostic system in Embodiment 1.

[0022] Figure 8 This is a flowchart illustrating an example of the diagnostic server decision processing in Embodiment 1.

[0023] Figure 9 This is a flowchart illustrating an example of the diagnostic server decision processing in Embodiment 1.

[0024] Figure 10 This is an example of AI selection information in Example 1.

[0025] Figure 11 This is an example of the data structure for the anonymized diagnostic object data in Example 1.

[0026] Figure 12 This is an example of a display screen showing the diagnostic results in Example 1.

[0027] Figure 13 This is an example of a display screen showing the diagnostic results in Example 1.

[0028] Figure 14 This is an example of a display screen showing the diagnostic results in Example 1.

[0029] Figure 15 This is an example of a display screen in Embodiment 1 showing the diagnostic results when image diagnostics are performed by multiple diagnostic servers.

[0030] Figure 16 This is an explanatory diagram showing a structural example of the image diagnostic system in Embodiment 2.

[0031] Figure 17 This is a timing diagram illustrating an example of image diagnostic processing in the image diagnostic system of Embodiment 2.

[0032] The reference numerals in the attached figures are explained as follows:

[0033] 100 Management Server, 101 Anonymization Processing Department, 102 AI Selection Department, 103 Display Screen Generation Department, 104 Diagnostic Result Data Generation Department, 110 AI Selection Information, 201 Diagnostic Server, 2011 Image Diagnostic Department, 300 In-Hospital Server, 301 Anonymization Processing Department, 302 Patient Information Management Department, 303 Display Screen Generation Department, 310 Patient Information Database, 400 Terminal, 401 Diagnostic Object Data Generation Department, 402 Additional Information Acquisition Department, 403 Display Screen Generation Department, 600 Computer, 601 Processor, 602 Storage Device, 603 Input Device, 604 Output Device, 605 Communication I / F, 900 Diagnostic Server, 2021 Learning Information Management Department, 2031 Diagnostic Image Generation Department, 2041 Management Department, 2101 Learning Database, 2111 Image Diagnostic Model. Detailed Implementation

[0034] Embodiments of the present invention will now be described with reference to the accompanying drawings. These embodiments are merely examples for implementing the present invention, and it should be noted that they do not limit the technical scope of the invention. Common structures are labeled with the same reference numerals in the various figures.

[0035] Example 1

[0036] Figure 1 This is an explanatory diagram illustrating the structure of an image diagnostic system according to Embodiment 1. The image diagnostic system includes a management server 100, a diagnostic server 201, a diagnostic server 202, and a diagnostic server 203. Additionally, the image diagnostic system includes an in-hospital server 300, a terminal 400, and an imaging device 500, all located in, for example, a hospital, clinic, or medical examination center. The in-hospital server 300, the terminal 400, and the imaging device 500 are connected via a network.

[0037] The imaging device 500 is an ophthalmic apparatus for imaging the fundus, and can include devices such as a fundus camera, a scanning laser ophthalmoscope, or an optical coherence tomography (OCT). The imaging device 500 is connected to the terminal 400. The imaging device 500 images the patient's examined eye, generating fundus image data for the right and left eyes. The generated fundus image data is then sent to the terminal 400.

[0038] In addition, fundus image data can be any one of the following: fundus images captured by a fundus camera, fundus images captured using a scanning laser ophthalmoscopy, and tomographic data of the fundus captured using an optical coherence tomography (OCT). Alternatively, it can be a combination of two or more of these, i.e., a fundus image dataset. Fundus image data is an example of image data of the eye being examined.

[0039] As an example of an image acquisition device, terminal 400 is a computer such as a PC (Personal Computer) or tablet used by doctors or operators of ophthalmic equipment. Terminal 400 is connected to an in-hospital server 300. Terminal 400 sends data, including fundus image data and additional information, as an example of first transmitted data to the in-hospital server 300.

[0040] In addition, the supplementary information includes one or a combination of the following: equipment information related to the performance and specifications of the imaging device 500; facility information including the department (ophthalmology, internal medicine, or diabetes medicine, etc.) of the hospital or clinic using the terminal 400, the cost of diagnostic procedures, and the name of the doctor; and diagnostic category information including the diagnostic mode and the name of the disease being diagnosed. Image attribute information, including the field of view, modality, and resolution of the image (image of the examined eye) captured by the imaging device 500, as well as the model and terminal ID of the imaging device 500, is an example of this equipment information. Modality refers to information indicating the type of imaging device 500 (e.g., fundus camera, scanning laser ophthalmoscope, optical coherence tomography, etc.) or the type of image captured by the imaging device 500 (e.g., fundus images or angiography images captured by red or near-infrared lasers, etc.). In addition, the names of doctors and hospitals using Terminal 400, the locations where terminals are installed (information on departments such as ophthalmology, internal medicine, and diabetes internal medicine, as well as information related to facilities such as optical shops or health check centers) are examples of facility information.

[0041] As an example of an image acquisition device, an in-hospital server 300 has a patient information database 310 for storing patient information received from the terminal 400. The in-hospital server 300 is connected to a management server 100 via a network. The in-hospital server 300 sends diagnostic subject data, as an example of first data transmission, to the management server 100. This diagnostic subject data includes patient information received from the terminal 400, fundus image data, and additional information. Furthermore, some or all of the patient information and some or all of the additional information in the diagnostic subject data can also be generated by the in-hospital server 300.

[0042] As an example of a first information processing device, a management server 100 generates anonymized diagnostic object data, as an example of a second data transmission. This anonymized diagnostic object data is derived by anonymizing a portion of the diagnostic object data (e.g., patient information) received from an in-hospital server 300. The management server 100 is connected to diagnostic servers 201, 202, and 203 via a network. Based on additional information, the management server 100 selects one of the diagnostic servers 201, 202, and 203 for image diagnosis of the fundus image data contained in the anonymized diagnostic object data, and sends the anonymized diagnostic object data to the selected diagnostic server.

[0043] Diagnostic servers 201-203, each serving as an example of an image diagnostic device, are equipped with AI (Artificial Intelligence) for image diagnosis of fundus image data. The functions (algorithms) of AI1, AI2, and AI3 equipped in diagnostic servers 201-203 are different (this will be described later). Upon receiving anonymized diagnostic data, the diagnostic server uses its equipped AI to perform image diagnosis on the fundus image data contained within the anonymized diagnostic data. The diagnostic results are encrypted and sent via management server 100 to in-hospital server 300 and terminal 400.

[0044] The following describes an example of each of the diagnostic servers 201-203. Here, the fundus image of the object and the diagnosed disease are used as examples for each AI, and various fundus images and diagnosed diseases can be combined.

[0045] As an example of an image diagnostic device, the diagnostic server 201 is a diagnostic server that takes fundus images captured by an imaging device 500 (with a narrow field of view, 30 to 100 degrees (less than) degrees) as an example of a first field of view, and is equipped with an AI 221 for diagnosing diabetic retinopathy. When the device information in the supplementary information indicates a narrow field of view, the management server 100 sends anonymized diagnostic object data to the diagnostic server 201.

[0046] The diagnostic server 202 is a diagnostic server that takes fundus images captured by an imaging device 500 (with the center of the eye as the starting point, the field of view is greater than or equal to 100 degrees and less than 200 degrees) or an ultra-wide-angle imaging device (with the center of the eye as the starting point, the field of view is greater than or equal to 200 degrees) as objects, and is equipped with an AI 222 for diagnosing diabetic retinopathy. When the device information in the supplementary information indicates a wide field of view and the diagnostic mode indicates diabetic retinopathy, the management server 100 sends anonymized diagnostic object data to the diagnostic server 202.

[0047] Furthermore, the diagnostic server 203 is a diagnostic server that takes fundus images captured by the ultra-wide-angle (with a field of view of ≥200 degrees, taking the center of the eyeball as the starting point) imaging device 500 as objects, and is equipped with an AI 223 capable of diagnosing not only diabetic retinopathy but also various fundus diseases. When the device information in the supplementary information indicates an ultra-wide-angle camera and the facility information indicates an ophthalmologist, the management server 100 sends anonymized diagnostic object data to the diagnostic server 203.

[0048] Figure 2This is a block diagram illustrating an example of the hardware structure of the computer that constitutes the management server 100, the diagnostic server, the in-hospital server 300, and the terminal 400. The computer 600, for example, has a processor (CPU) 601, a storage device 602, an input device 603, an output device 604, and a communication I / F (Interface) 605, which are interconnected via internal signal lines 606.

[0049] Processor 601 executes a program stored in storage device 602. Storage device 602 includes memory. Memory includes ROM, which is a non-volatile storage element, and RAM, which is a volatile storage element. ROM stores immutable programs (such as BIOS). RAM is a high-speed and volatile storage element such as DRAM (Dynamic Random Access Memory), which temporarily stores the program executed by processor 601 and the data used during program execution.

[0050] Additionally, storage device 602 includes an auxiliary storage device. This auxiliary storage device is, for example, a high-capacity, non-volatile storage device such as a magnetic storage device (HDD) or a flash memory (SSD), which stores the program executed by processor 601 and the data used during program execution. That is, the program is read from the auxiliary storage device, loaded into memory, and executed by processor 601.

[0051] Input device 603 is an interface that accepts input from the operator, such as a keyboard or mouse. Output device 604 is a device that outputs the execution results of the program in a form that the operator can visually confirm, such as a monitor or printer. Alternatively, input device 603 and output device 604 can be integrated, such as a touch panel device. Communication I / F 605 is a network interface device that controls communication with other devices in accordance with a specified protocol.

[0052] The program executed by the processor 601 is provided to the computer 600 via removable media (CD-ROM, flash memory, etc.) or a network and stored in a non-volatile auxiliary storage device that serves as a non-temporary storage medium. Therefore, the computer 600 can have an interface for reading data from the removable media.

[0053] In addition, the management server 100, the diagnostic server, the in-hospital server 300 and the terminal 400 are computer systems that are physically located on one computer 600 or are logically or physically located on multiple computers 600. They can work on the same computer 600 with different threads, or they can work on a hypothetical computer built on multiple physical computer resources.

[0054] Figure 3This is a block diagram illustrating an example of the functional structure of the management server 100. The management server 100 includes an anonymization processing unit 101, an AI selection unit 102, a display screen generation unit 103, and a diagnostic result data generation unit 104. The anonymization processing unit 101 anonymizes the patient information contained in the diagnostic object data sent from the hospital server 300. The AI ​​selection unit 102 selects an AI for image diagnosis based on the additional information contained in the diagnostic object data, targeting the fundus image data contained in the diagnostic object data.

[0055] The display screen generation unit 103 generates screen information to be displayed on the output device 604. The diagnostic result data generation unit 104 decrypts the encrypted diagnostic results received from the diagnostic server and generates a display screen showing the diagnostic results. Figure 12 , Figure 13 , Figure 14 The information displayed on the screen is then sent to the hospital's server 300.

[0056] Furthermore, the functional units included in the management server 100 are implemented by the processor 601 of the computer 600 that implements the management server 100. Specifically, the processor 601 functions as an anonymization processing unit 101 by operating in accordance with an anonymization processing program loaded in the memory included in the storage device 602, and functions as an AI selection unit 102 by operating in accordance with an AI selection program loaded in the memory included in the storage device 602. Similarly, other functional units included in the management server 100 and functional units included in other devices are also implemented by the processor 601 operating in accordance with programs loaded in memory.

[0057] Management server 100 maintains AI selection information 110. AI selection information 110 maintains the correspondence between additional information and diagnostic servers 201, 202, and 203. Furthermore, as will be described later, since AI 221, AI 222, and AI 223 each contain different image diagnostic models, it can also be said that AI selection information 110 contains the correspondence between additional information and image diagnostic models. In AI selection information 110, AIs corresponding to additional information are determined by describing conditional branches based on the values ​​of one or more additional information pieces. Additionally, in AI selection information 110, the correspondence between the values ​​(or ranges of values) of one or more additional information pieces and AI 220 can also be described in tabular form.

[0058] AI selection information 110 is stored in an auxiliary storage device included in the storage device 602 of the computer 600 that implements the management server 100. Similarly, information and databases held by other devices are also stored in the auxiliary storage device included in the storage device 602 of the computer 600 that implements those other devices.

[0059] Furthermore, in this embodiment, the information used by the various devices included in the image diagnostic system is not dependent on a data structure, and can be represented by any data structure. For example, a data structure appropriately selected from a table, list, database, or queue can store the information. All information is stored in non-volatile memory, etc.

[0060] Figure 4 This is a block diagram illustrating an example of the functional structure of diagnostic server 201. Diagnostic servers 201, 202, and 203 differ only in their AI functions; other structures (such as the display generation unit and management unit, which perform the same functions) are identical. Therefore, the functional structure of diagnostic server 201 will be explained.

[0061] The diagnostic server 201 includes, for example, an image diagnostic unit 2011, a learning information management unit 2021, a diagnostic image generation unit 2031, and a management unit 2041. Furthermore, the diagnostic server 201 maintains a learning database 2101 and an image diagnostic model 2111. The learning database 2101 is a database used to construct the image diagnostic model 2111. The image diagnostic model 2111 is a model that outputs diagnostic results if image data is input; it takes fundus images captured by an imaging device 500 with a narrow field of view (a field of view of 30 to 100 degrees (less than) degrees) as the object and outputs the severity of diabetic retinopathy symptoms as the diagnostic result. In this embodiment, the severity of diabetic retinopathy symptoms is set to a five-stage International Severity Classification (ISC).

[0062] AI221 is implemented through the image diagnostic unit 2011, the learning information management unit 2021, the learning DB 2101, and the image diagnostic model 2111. The image diagnostic unit 2011 performs image diagnosis using the image diagnostic model 2111 on the fundus image data contained in the anonymized diagnostic object data received from the management server 100.

[0063] In 2021, the Learning Information Management Department stored the examined eye image data and image diagnosis results contained in the anonymized diagnostic object data as learning data for AI in the Learning DB210 and updated the Learning DB210. The Learning Information Management Department in 2021 updated (e.g., optimized) the image diagnosis model 2111 by learning based on the updated Learning DB2101.

[0064] The diagnostic image generation unit 2031 generates a completed diagnostic fundus image by overlaying markers indicating the location of the lesion and / or text indicating the disease name onto the diagnosed fundus image. The management unit 2041 manages the AI ​​221. The completed diagnostic fundus image, along with the diagnostic results, is sent to the management server 100.

[0065] Furthermore, the diagnostic server 201 may not have the learning function of the image diagnostic model 2111. That is, the diagnostic server 201 may not update the image diagnostic model 2111 but instead continue to perform image diagnosis with the pre-determined image diagnostic model 2111 fixed. In this case, the diagnostic server 201 may also not have the learning information management unit 2021 and the learning DB 2101.

[0066] Diagnostic servers 202 and 203 have different image diagnostic models than diagnostic server 201, but apart from this aspect, they have the same structure as diagnostic server 201.

[0067] The image diagnostic model maintained by the diagnostic server 202 is a model that outputs diagnostic results if image data is input. It takes fundus images captured by the imaging device 500 with a wide field of view (the field of view is greater than or equal to 100 degrees and less than 200 degrees with the center of the eyeball as the starting point) or an ultra-wide field of view (the field of view is greater than or equal to 200 degrees with the center of the eyeball as the starting point) as the object and outputs the symptom level of diabetic retinopathy as the diagnostic result.

[0068] The image diagnostic model maintained by the diagnostic server 203 is a model that outputs diagnostic results if image data is input. It takes fundus images captured by the ultra-wide-angle (with the eye center as the starting point and the field of view greater than or equal to 200 degrees) imaging device 500 as objects and outputs diagnostic results not only for diabetic retinopathy but also for various fundus diseases.

[0069] Furthermore, the image diagnostic department, learning information management department, and learning DB of diagnostic server 202 and diagnostic server 203 are adapted to the image diagnostic model they maintain.

[0070] Figure 5 This is a block diagram illustrating an example of the functional structure of an in-hospital server 300. The in-hospital server 300 includes, for example, an anonymization processing unit 301, a patient information management unit 302, and a display screen generation unit 303. Additionally, the in-hospital server 300 maintains patient information DB 310.

[0071] The anonymization processing unit 301 anonymizes the patient information contained in the diagnostic data. The patient information management unit 302 stores the patient information contained in the diagnostic data in the patient information DB 310, retrieves patient information from the patient information DB 310, and appends it to the diagnostic data. The display screen generation unit 303 generates screen information to be displayed on the output device 604. The patient information DB 310 stores the patient's information.

[0072] Figure 6This is a block diagram illustrating an example of the functional structure of terminal 400. Terminal 400 includes a diagnostic object data generation unit 401, an additional information acquisition unit 402, and a display screen generation unit 403. The diagnostic object data generation unit 401 generates diagnostic object data including patient information, additional information, and image data of the examined eye. The additional information acquisition unit 402 acquires additional information for selecting an AI (or a diagnostic server with an AI suitable for diagnosis). The display screen generation unit 403 generates screen information to be displayed on the output device 604.

[0073] Figure 7 This is a timing diagram illustrating the image diagnostic processing of the image diagnostic system in Embodiment 1. Figure 7 In the example, a diagnostic server for image diagnosis of fundus images is selected based on device information such as the imaging device 500.

[0074] First, the diagnostic data generation unit 401 of terminal 400 accepts the input of patient information via input device 603 (S701). Patient ID, age, gender, address, medical history, medication history, and consultation results are examples of patient information. Furthermore, for patients whose information has already been entered into patient information DB310, accepting the input of the patient ID eliminates the need to input other patient information.

[0075] The diagnostic subject data generation unit 401 acquires fundus image data of both eyes of the patient sent from the imaging device 500 (S702). Furthermore, in this embodiment, fundus image data of both eyes can be acquired, or fundus image data of only the left or right eye can be acquired. The diagnostic subject data generation unit 401 generates left and right eye markers indicating whether the fundus image data is data of both eyes, data of only the right eye, or data of only the left eye. The diagnostic subject data generation unit 401 can also acquire fundus image data from a device other than the imaging device 500.

[0076] Next, the additional information acquisition unit 402 acquires additional information (S703). Specifically, for example, the additional information acquisition unit 402 acquires device information from the shooting device 500 as additional information, or accepts input from hospitals or doctors' clinics using the terminal 400.

[0077] Furthermore, this additional information can also be pre-stored in the storage device 602 of the terminal 400. Alternatively, for example, the imaging device 500 can embed device information as metadata into the fundus image data, and the additional information acquisition unit 402 can acquire the device information from the fundus image data.

[0078] Next, the diagnostic subject data generation unit 401 sends the diagnostic subject data, which includes patient information, fundus image data, left and right eye markers, and additional information, to the management server 100 via the hospital server 300 (S704). In addition, the patient information management unit 302 of the hospital server 300 saves the patient information received from the terminal 400 in the patient information DB310.

[0079] Furthermore, if the patient information received from the terminal 400 is incomplete, the patient information management unit 302 can supplement it by referring to the patient information DB310 to obtain patient information. Specifically, for example, if the patient information received from the terminal 400 is only the patient's ID, the patient information management unit 302 obtains the patient information corresponding to that ID from the patient information DB310 and sends the diagnostic data, including the obtained patient information, to the management server 100.

[0080] Next, the anonymization processing unit 101 of the management server 100 performs anonymization processing based on a prescribed algorithm on the patient information contained in the received diagnostic data (S705). Anonymization processing includes anonymizing the patient ID (replacing it with the ID inherent in the fundus image data) or deleting personal information such as the patient's name and disease name. Furthermore, the anonymization processing unit 101 may also anonymize only a portion of the patient information (e.g., only sensitive information related to privacy). Additionally, the anonymization processing of patient information may be pre-executed, for example, by the anonymization processing unit 301 of the hospital server 300 before sending the diagnostic data to the management server 100.

[0081] The AI ​​selection unit 102 of the management server 100 selects at least one from the diagnostic servers 201, 202, and 203 based on the AI ​​selection information 110 and the additional information contained in the received diagnostic subject data. It then selects the anonymized diagnostic subject data (see below) containing anonymized patient information, fundus image data, left and right eye markers, and additional information. Figure 11 (S706) Send the message to the selected diagnostic server (here, diagnostic server 201 is selected). Details of step S706 will be described later.

[0082] In addition, the AI ​​selection unit 102 may also send the anonymized diagnostic object data to the selected diagnostic server 201 after encrypting it with an encryption key. In this case, the diagnostic server 201 has a decryption key corresponding to the encryption key, and the anonymized diagnostic object data is first decrypted using the decryption key in step S707 described later.

[0083] Next, the image diagnosis unit 2011 of the diagnosis server 201, which receives the anonymized diagnostic object data, uses the image diagnosis model 2111, which diagnoses diabetic retinopathy based on narrow-angle fundus images, to perform image diagnosis on the fundus image data contained in the received anonymized diagnostic object data (S707).

[0084] The learning information management unit 2021 of the diagnostic server 201 updates the learning DB 2101 by storing fundus image data as learning data in the learning DB 2101, and updates the image diagnostic model 2111 based on the updated learning DB 2101 (S708). In addition, the learning information management unit 2021 may also store anonymized patient information and additional information together as learning data in the learning DB 2101.

[0085] Next, the management unit 2041 generates image diagnostic result data that includes at least anonymized patient information and image diagnostic results. Using the encryption key held by the diagnostic server 201, the image diagnostic results are encrypted, and an encrypted image diagnostic result is generated and sent to the management server 100 (S709). The image diagnostic result data may also include a diagnosed fundus image obtained by overlaying markers indicating the location of the lesion and / or text indicating the disease name onto the diagnosed fundus image.

[0086] Next, the diagnostic result data generation unit 104 of the management server 100 decrypts the received encrypted image diagnostic result using the decryption key held by the management server 100 (S710). Furthermore, the diagnostic result data generation unit 104 restores the anonymized patient information. It then establishes an association between the decrypted image diagnostic result and the patient information of the patient before anonymization.

[0087] Next, the diagnostic result data generation unit 104 generates a display screen showing the diagnostic result, including the grade of diabetic retinopathy as a diagnostic result. Figure 12 (and then assign a patient ID and save it in a memory not shown).

[0088] When the management server 100 receives the encrypted image diagnostic result from the diagnostic server 202, the diagnostic result data generation unit 104 generates a display screen showing the diagnostic result. Figure 13 When the management server 100 receives the encrypted image diagnostic result from the diagnostic server 203, the diagnostic result data generation unit 104 generates a display screen showing the diagnostic result. Figure 14 ).

[0089] Next, the diagnosis result data generation unit 104 associates the display screen representing the diagnosis result with the patient information and sends it to the hospital server 300 (S711). The display screen generation unit 303 of the hospital server 300 displays the display screen representing the received diagnosis result and the patient information on the output device 604 of the hospital server 300 (S712). Alternatively, the terminal 400 may obtain the image diagnosis result and the patient information from the hospital server 300, and the display screen generation unit 403 of the terminal 400 may display the display screen representing the diagnosis result and the patient information on the output device 604 of the terminal 400.

[0090] Alternatively, the display screen generation unit 103 of the management server 100 may generate display screen information based on the image diagnosis results and patient information and send the information to the hospital server 300. The hospital server 300 and the terminal 400 shall display the display screen in accordance with the generated information.

[0091] Figure 8 This is a flowchart illustrating an example of the diagnostic server's decision to process step S706. Figure 8 In the example, the diagnostic server decides to send anonymized diagnostic object data based on the field of view information contained in the supplementary information. First, the AI ​​selection unit 102 of the management server 100 obtains the supplementary information containing the field of view information from the diagnostic object data (S801).

[0092] The AI ​​selection unit 102 determines whether the field of view shown in the field of view information is a wide field of view (S802). Specifically, when the field of view information shows a specific angle, if the angle is a predetermined value (e.g., 100 degrees) or more, it is determined to be a wide field of view; if it is less than the predetermined value, it is determined not to be a wide field of view (it is a narrow field of view).

[0093] Alternatively, the management server 100 may also maintain device information for each shooting device 500. Specifically, this device information is a lookup table defined by establishing correspondences between the model, terminal ID, field of view, and resolution of each shooting device 500. In this case, the AI ​​selection unit 102 may also obtain the model or terminal ID of the shooting device 500 from the supplementary information and, referring to the device information, determine whether the field of view of the shooting device 500 indicated by the model or terminal ID is a wide field of view.

[0094] If the AI ​​selection unit 102 determines that the field of view shown in the field of view information is a wide field of view (S802: Yes), it sends anonymized diagnostic object data to the diagnostic server 202 (S803). If it determines that the field of view is not a wide field of view (S802: No), it sends anonymized diagnostic object data to the diagnostic server 201 (S804), and ends the processing in step S705.

[0095] exist Figure 8 In this configuration, the AI ​​mounted on diagnostic server 202 is designed to perform high-precision image diagnosis on fundus images captured with a wide field of view, while the AI ​​mounted on diagnostic server 201 is designed to perform high-precision image diagnosis on images of the examined eye captured with a narrow field of view. Therefore, the AI ​​selection unit 102 can select the appropriate diagnostic server based on the field of view of the image of the examined eye.

[0096] In addition, Figure 8 In this process, the AI ​​selection unit 102 determines whether the field of view shown in the field of view information is a wide field of view, and selects the diagnostic server for the destination from the two diagnostic servers 201 and diagnostic server 202. Alternatively, the AI ​​selection unit 102 can also determine which interval of three or more field of view intervals the field of view shown in the field of view information matches, and select the diagnostic server corresponding to that interval by referring to the AI ​​selection information 110.

[0097] In addition, Figure 8 While the diagnostic server is selected based on the field of view, other elements included in the supplementary information can also be used. For example, facility information included in the supplementary information (such as the names of doctors and hospitals on terminal 400, the locations where terminal 400 is installed (information on departments such as ophthalmology, internal medicine, and diabetes internal medicine, as well as information related to facilities such as optical shops and health check centers), and the fees for diagnostic items in hospitals or clinics where terminal 400 is installed) can be used by the AI ​​selection unit 102 to determine the diagnostic server.

[0098] At this time, for example, if the department information is ophthalmology, the AI ​​selection unit 102 selects diagnosis server 203; if the department information is internal medicine, it selects diagnosis server 201. Additionally, for example, if the cost of the diagnostic item is above a specified value, the AI ​​selection unit 102 selects diagnosis server 202 or diagnosis server 203; if the cost of the diagnostic item is below the specified value, it selects diagnosis server 201.

[0099] Furthermore, the system includes a diagnostic server 201 equipped with AI capable of accurately diagnosing the symptoms of a specific disease (e.g., diabetic retinopathy) and a diagnostic server 203 capable of comprehensively diagnosing the symptoms of multiple diseases. Assume that the supplementary information includes information indicating the disease being diagnosed. In this case, for example, if the disease of the diagnostic object indicated by the supplementary information is that specific disease, the AI ​​selection unit 102 selects the diagnostic server 201; if the disease of the diagnostic object indicated by the supplementary information is not a specific disease but a comprehensive diagnosis, then the diagnostic server 203 is selected.

[0100] Furthermore, for example, a diagnostic server 201 equipped with AI capable of performing high-precision image diagnosis on images of the examined eye captured by a fundus camera, and a diagnostic server 202 equipped with AI capable of performing high-precision image diagnosis on images of the examined eye captured by a scanning laser ophthalmoscope. Assume that the supplementary information includes information on the representation format. In this case, for example, if the form shown in the supplementary information is a fundus camera, the AI ​​selection unit 102 selects diagnostic server 201; if the form shown in the supplementary information is a scanning laser ophthalmoscope, it selects diagnostic server 202.

[0101] In this way, the AI ​​selection unit 102 can select the diagnostic server equipped with the most suitable AI from multiple diagnostic servers based on additional information. As a result, users will not hesitate or be troubled by the selection of a diagnostic server and can send fundus image data to the appropriate diagnostic server.

[0102] Figure 9 This is a flowchart illustrating other examples of the diagnostic server's decision to process step S706. Figure 9 In this context, the diagnostic server that sends anonymized diagnostic data is determined based on the various information contained in the supplementary information (field of view information and departmental information). In other words, the field of view information and departmental information are assumed to be included in the supplementary information. (Explanation and...) Figure 8 The differences.

[0103] If the AI ​​selection unit 102 determines that the field of view shown in the field of view information is a wide field of view (S802: Yes), it determines whether the department shown in the department information is ophthalmology or internal medicine (S901). If the AI ​​selection unit 102 determines that the department shown in the department information is ophthalmology (S901: ophthalmology), it sends anonymized diagnostic object data to the diagnostic server 203 (S803); if it determines that the department is internal medicine (S901: internal medicine), it sends anonymized diagnostic object data to the diagnostic server 202 (S902), and the processing in step S705 ends.

[0104] exist Figure 9 The AI ​​mounted on diagnostic server 202 is designed to perform detailed diagnosis of specific diseases (such as diabetic retinopathy) with high precision based on fundus image data captured with a wide field of view, including symptom severity. Additionally, the AI ​​mounted on diagnostic server 201 is designed to perform image diagnosis with high precision based on fundus image data captured with a narrow field of view. Furthermore, the AI ​​mounted on diagnostic server 203 is designed to diagnose the symptoms of various diseases with high precision based on fundus image data captured with a wide field of view.

[0105] Therefore, the AI ​​selection unit 102 can select an appropriate diagnostic server based on the field of view of the image of the examined eye and the hospital where the terminal 400 is installed or the department of the doctor using the terminal 400. In other words, it can send fundus image data to the appropriate diagnostic server based on various information (field of view information and department information) included in the supplementary information.

[0106] In addition, although Figure 9 In the process, all fundus image data captured with a narrow field of view is sent to the diagnostic server 201. However, for the image data of the examined eye captured with a narrow field of view, a conditional branch based on additional information can be added to select the diagnostic server as the destination from multiple different diagnostic servers.

[0107] In addition, although Figure 9 The method uses two types of additional information to select a specific diagnostic server from multiple diagnostic servers, but the mapping between additional information and diagnostic servers can be arbitrarily designed based on the characteristics of the AI ​​mounted on each diagnostic server. For example, more than three types of additional information can also be used, and the diagnostic server can be determined by using arbitrary conditional branches based on multiple types of additional information.

[0108] Figure 10 This is an example of AI selecting information 110. Figure 10 In this example, the correspondence between the supplementary information and the diagnostic server is described in tabular form. AI selection information 110 includes, for example, a record number field 1101, a supplementary information field 1102, and a diagnostic server ID field 1103. The record number field 1101 stores the record number that identifies the AI ​​selection information 110. The supplementary information field 1102 stores elements of one or more supplementary information types (field of view information, form information, department information, resolution information, etc.). The diagnostic server ID field 1103 stores the ID of the destination server for the anonymized diagnostic data corresponding to the combination of supplementary information.

[0109] For example, in step S706, the AI ​​selection unit 102 obtains additional information and directs the selection to the user with... Figure 10 The diagnostic server corresponding to the diagnostic server ID in the table with the obtained additional information sends anonymized diagnostic object data. Furthermore, Figure 8 and Figure 9 Such conditional branches can also be passed Figure 10 That kind of tabular form of description.

[0110] Figure 11This is an example of a data structure for anonymized diagnostic object data sent from management server 100 to diagnostic server. Anonymized diagnostic object data may include, for example, head data 701, DICOM (Digital Imaging and Communications in Medicine) data 702, user data 703, and fundus image data 704.

[0111] The header 701 contains information such as the data source and destination, and the data type (medical images, documents, emails, etc.). The DICOM data 702 contains information such as the format of medical images captured by the imaging device 500, and information defining the communication protocol between medical devices including the imaging device 500.

[0112] User data 703 includes, for example, a diagnostic category flag, left / right eye flags, anonymized patient information, and additional information. The diagnostic category flag indicates the name of the disease diagnosed by the diagnostic server selected by the AI ​​selection unit 102 (a value identifying diabetic retinopathy, age-related macular degeneration, and all fundus diseases, etc.). The left / right eye flag indicates whether the fundus image data 704 is image data for the right eye, the left eye, or both eyes (e.g., a value among L, R, or LR). Furthermore, information such as the terminal ID and format of the imaging device 500 in the additional information can also be recorded in the DICOM data 702, or it can be recorded only in the DICOM data 702.

[0113] Figure 12 This is an example of a display screen showing the diagnostic results based on the diagnostic server 201. Figure 12 This is the display screen (screen layout) used to diagnose the symptom level of diabetic retinopathy based on fundus images (images of the examined eye) taken with a narrow field of view.

[0114] The display screen includes a patient information display area 1201, an AI information display area 1202, a supplementary information display area 1203, and a diagnosis result display area 1204. The patient information display area 1201 displays, for example, patient information contained in the image diagnostic data. The AI ​​information display area 1202 displays, for example, the ID of the diagnostic server that performed the image diagnosis and the ID (or version number) of the AI ​​that performed the image diagnosis. The supplementary information display area 1203 displays, for example, part or all of the supplementary information contained in the diagnostic object data (field of view, resolution, diagnostic category, etc.).

[0115] The diagnostic results display area 1204 shows information about the diagnostic results based on the fundus image data diagnosed by the diagnostic server. Figure 12 In the example, fundus images of both eyes, rectangular bars showing the symptom levels of the five stages of diabetic retinopathy in both eyes, and observations of both eyes are displayed in the diagnostic results display area 1204.

[0116] exist Figure 12 In the example, the right eye image and a right-pointing arrow (indicator) showing the symptom level of diabetic retinopathy in the right eye are displayed on the right side of the rectangular bar, while the left eye image and a left-pointing arrow showing the symptom level of diabetic retinopathy in the left eye are displayed on the left side of the rectangular bar. Thus, a user can grasp the symptom levels of diabetic retinopathy in both eyes and the differences in symptom levels with just a glance at the diagnostic results display area 1204. Furthermore, the observations in the diagnostic results display area 1204 are generated by the AI ​​of the diagnostic server. Alternatively, the observations can be input and edited by the user of the terminal 400.

[0117] Figure 13 This is an example of a display screen showing the diagnostic results based on the diagnostic server 202. Figure 13 This is a display showing a diagnosis of the symptom level of diabetic retinopathy based on wide-angle fundus images captured with a wide field of view. In diabetic retinopathy, early symptoms manifest as abnormalities in the peripheral region of the fundus, which then spread towards the center. Capturing the fundus with a wide field of view yields wide-angle fundus images that include not only the central but also the peripheral regions. Therefore, AI performing image diagnosis using wide-angle fundus images can not only predict the current symptom level of diabetic retinopathy but also predict future symptom levels considering the condition of the peripheral fundus.

[0118] Therefore, in Figure 13 In the example, the diagnostic results display area 1204 shows not only rectangular bars indicating the current symptom levels of the five stages of diabetic retinopathy in both eyes, but also rectangular bars indicating the future symptom levels of diabetic retinopathy in both eyes. Furthermore, the observation results in the diagnostic results display area 1204 also record future (in...) Figure 13 The prediction is set to one year later, but it can also be three months later, six months later, etc.

[0119] In addition, although Figure 13 In the example, the current symptom level and the future symptom level are displayed in the diagnosis result display area 1204, but the current symptom level in the central part of the fundus and the current symptom level in the peripheral part of the fundus can also be displayed in the diagnosis result display area 1204.

[0120] Figure 14This is an example of a display screen showing the diagnostic results based on the diagnostic server 203. Figure 14 This is a display screen shown when diagnosing various diseases using wide-angle fundus images captured with a wide field of view. One of these diseases is diabetic retinopathy. This is related to... Figure 13 The diagnostic results for diabetic retinopathy showed the same outcome. Figure 14 The text further displays rectangular bars and arrows (indicators) showing the category and severity of symptoms for diseases 2 and 3. Diseases 2 and 3 are fundus diseases such as age-related macular degeneration and retinal detachment that can be identified by the AI ​​223 of the diagnostic server 203.

[0121] The following describes a variation of the display screen showing the diagnostic results.

[0122] In cases where an image diagnosis is performed that can diagnose the presence or absence of symptoms but cannot determine the severity of the symptoms, information about the disease with inferred symptoms can be displayed in the diagnosis result display area 1204.

[0123] In this case, the diagnostic results display area 1204 can further display a message recommending image diagnosis based on an AI with specific symptom levels that can diagnose the inferred symptoms of the disease.

[0124] Additionally, the management server 100 can maintain correspondence information between diseases and diagnostic servers equipped with AI capable of diagnosing the symptom levels of those diseases. In this case, if the management server 100 receives a command from the hospital server 300 to perform image diagnosis according to the message, it refers to the correspondence information and determines a diagnostic server equipped with AI capable of diagnosing the symptom levels of the disease from among the diagnostic servers capable of sending the command. Furthermore, the management server 100 can send information representing that diagnostic server to the hospital server 300 and display it, or it can resend anonymized diagnostic data to that diagnostic server and entrust image diagnosis to it.

[0125] Additionally, the image diagnostic system is configured to include a diagnostic server equipped with an AI (AI A) for image diagnosis of low-resolution images (e.g., a first resolution less than a predetermined value), and a diagnostic server equipped with an AI (AI B) for image diagnosis of high-resolution images (e.g., a second resolution greater than or equal to the predetermined value). In this case, if the diagnostic result based on AIA indicates an abnormality in the fundus, a message recommending image diagnosis based on AIA B using higher-resolution fundus image data can be displayed in the diagnostic result display area 1204. Alternatively, a message recommending taking images using higher-resolution fundus image data can also be displayed.

[0126] In addition, anonymized diagnostic data can be sent to multiple diagnostic servers, meaning that image diagnosis of fundus image data can be performed using AIs with different image diagnostic models. Figure 15 This is an example of a display screen showing the diagnostic results when image diagnostics are performed through multiple diagnostic servers. Figure 15 This is an example of a display showing a scenario where the diagnosis of symptom severity in diabetic retinopathy based on wide-angle fundus images captured with a wide field of view is performed by two diagnostic servers. Furthermore, it is assumed that the symptom severity classifications and the number of classifications shown in the diagnostic results from the two diagnostic servers differ.

[0127] In this case, for example, Figure 15 The diagnostic results display area 1204 shows a first diagnostic result and a second diagnostic result. The first diagnostic result was determined by a diagnostic server equipped with AI that performs image diagnosis of diabetic retinopathy based on Category 1 (the International Classification of Disease, which classifies diabetic retinopathy into five stages). The second diagnostic result was determined by a diagnostic server equipped with AI that performs image diagnosis of diabetic retinopathy based on Category 2 (the Davis Classification, which classifies diabetic retinopathy into three stages). In other words, it displays the symptom levels of diabetic retinopathy based on two different classifications.

[0128] Alternatively, the display generation unit 103 of the management server 100 can also generate symptom levels that take into account symptom levels based on different classifications and display them as a comprehensive result.

[0129] Specifically, the display screen generation unit 103 predetermines the scores corresponding to the symptom levels of each category (for example, No DR in category 1 is 1 point, Mild is 3 points, ..., A1 in category 2 is 1 point, A2 is 7 points, ... etc.), and calculates the average score of the left and right eyes corresponding to the symptom levels.

[0130] Furthermore, in the overall result, a symptom level corresponding to each interval of the average score is predetermined (symptom level 1 is greater than or equal to 1 point and less than 2.5 points, symptom level 2 is greater than or equal to 2.5 points and less than 5 points, etc.). The display screen generation unit 103 determines the symptom level corresponding to the calculated average score for each eye pair and displays the determined symptom level as the overall result in the diagnosis result display area 1204. This allows for the integration of diagnostic results from AIs diagnosing diabetic retinopathy based on multiple different classifications to suggest new indicators. Furthermore, classification 2 is not limited to the Davis classification (modified), but can also be the Fukuda classification, etc.

[0131] Furthermore, although in this embodiment the management server 100 determines the diagnostic server for image diagnosis of the fundus image data included in the anonymized diagnostic object data, other devices (such as the in-hospital server 300, terminal 400, or imaging device 500) can also determine the diagnostic server. In this case, the other device maintains the AI ​​selection information 110. Additionally, the other device can include information indicating the determined diagnostic server (e.g., a flag) in the user data 703.

[0132] Example 2

[0133] The diagnostic server 900 in the image diagnostic system of Example 2 is equipped with multiple AIs 901-902. The differences from Example 1 will be explained.

[0134] Figure 16 This is an explanatory diagram showing a structural example of the image diagnostic system of Embodiment 2. Regarding the diagnostic server 900 equipping with multiple AIs, and... Figure 1 The image diagnostic system differs from this one. Furthermore, to simplify the explanation, in Figure 16 The image diagnostic system has one diagnostic server 900, but it can also have multiple diagnostic servers, and some or all of these multiple diagnostic servers are equipped with multiple AIs. The multiple AIs (including two or more AIs such as AI901 and AI902) each have different image diagnostic models.

[0135] The AI ​​selection information 110 in Example 2 includes information used to determine the AI ​​using additional information sent from the in-house server 300 or terminal 400. That is, for example, in... Figure 10 In the AI ​​selection information 110, there may also be an AI ID field. The AI ​​ID is the ID used to identify the AI ​​performing image diagnosis on the anonymized diagnostic object data corresponding to the value of the additional information.

[0136] Figure 17 This is a timing diagram illustrating the image diagnostic processing of the image diagnostic system in Embodiment 2. (Explanation and...) Figure 7 The difference is that step S1601 is performed instead of step S702. In step S1601, the AI ​​selection unit 102 of the management server 100 selects at least one AI (two or more AIs including AI901 and AI902) based on the AI ​​selection information 110 and the additional information contained in the received diagnostic object data.

[0137] Furthermore, in step S1601, the AI ​​selection unit 102 includes the AI ​​information (such as the AI's ID) representing the selected AI 220 in the user data 703, and sends the anonymized diagnostic object data containing the user data 703 to the diagnostic server equipped with the selected AI 220.

[0138] Next, the management unit of the diagnostic server that received the anonymized diagnostic object data selected AI220, which is the AI ​​information contained in the anonymized diagnostic object data, and input the fundus image data contained in the anonymized diagnostic object data into the selected AI220 (S1602). Then, the process proceeds to step S707.

[0139] Furthermore, although in Embodiment 2 the AI ​​for image diagnosis of the fundus image data included in the anonymized diagnostic subject data is determined by the management server 100, the AI ​​can also be determined by other devices (e.g., diagnostic server 900, in-hospital server 300, terminal 400, or imaging device 500). In this case, the other device retains the AI ​​selection information 110. Additionally, the other device includes information indicating the determined AI (e.g., a flag) in the user data 703.

[0140] However, when diagnostic server 900 decides on the AI, since management server 100 cannot determine which diagnostic server 900 maintains the appropriate AI, it expects to send the anonymized diagnostic data to all diagnostic servers. Furthermore, upon receiving the anonymized diagnostic data, the diagnostic server 900 refers to the AI ​​selection information 110 to determine whether it can perform image diagnosis of the fundus image data using its own AI. And, if the diagnostic server is equipped with AI capable of performing image diagnosis, it sends the diagnostic results based on the AI ​​diagnosis to management server 100. Furthermore, the present invention is not limited to the above-described contents, and these contents can be combined in any way. Additionally, other embodiments conceivable within the scope of the technical concept of the present invention are also included within the scope of the present invention.

Claims

1. An information processing system, characterized in that, have: Image acquisition device for acquiring image data of the patient's examined eye; and A first information processing device capable of communicating with the image acquisition device and storing the image data of the examined eye. The image acquisition device performs the first transmission process. The first transmission process sends first transmission data, including the image data of the examined eye and additional information for determining the image diagnostic device for performing image diagnosis on the image data of the examined eye, to the first information processing device. The first information processing device performs the following processing: Storage processing: If the first transmitted data is received from the image acquisition device, the image data of the examined eye is stored. The process involves determining at least one of a first image diagnostic device and a second image diagnostic device based on the additional information, wherein the first image diagnostic device performs a first image diagnosis on the examined eye image data, and the second image diagnostic device performs a second image diagnosis on the examined eye image data that differs from the first image diagnosis; and The second transmission process involves sending second transmission data, containing the image data of the examined eye, to the identified image diagnostic device. The additional information is determined based on the attribute information of the examined eye image data. The attribute information includes field-of-view information, which indicates the field of view of the image captured when the image data of the examined eye was taken. In the first image diagnosis, image diagnosis is performed on the image data of the examined eye captured at the first field of view. In the second image diagnosis, image diagnosis is performed on the image data of the examined eye taken at a second field of view that is wider than the first field of view.

2. The information processing system as described in claim 1, characterized in that, If the first image diagnostic device detects an abnormality in the examined eye image data during the first image diagnosis, it will include information indicating a recommendation to acquire image data of the examined eye taken at the second field of view in the diagnostic result of the image diagnosis. The first image diagnostic device sends the diagnostic results to the first information processing device.

3. The information processing system as described in claim 1, characterized in that, The image acquisition device performs the following processing: The process involves acquiring the attribute information; and The decision is made to process the information based on the acquired attribute information, determining the flag of at least one of the first image diagnostic device and the second image diagnostic device as the additional information. In the first transmission process, the flag is sent to the first information processing device.

4. The information processing system as described in claim 1, characterized in that, The image acquisition device performs an acquisition process that uses the attribute information of the patient's examined eye image data as additional information. The first information processing device performs the determination process based on the received attribute information.

5. The information processing system as described in claim 1, characterized in that, The examined eye image data includes at least one of fundus image data obtained from a fundus camera, fundus image data obtained from a scanning laser ophthalmoscope, and tomographic data obtained from an optical coherence tomography (OCT) system.

6. The information processing system as described in claim 5, characterized in that, The examined eye image data includes fundus image data. In the first image diagnosis and the second image diagnosis, the diagnosis of fundus lesions is performed.

7. The information processing system as described in claim 6, characterized in that, In the first image diagnosis and the second image diagnosis, a diagnosis of diabetic retinopathy was performed using fundus images.

8. The information processing system as described in claim 6, characterized in that, In the first image diagnosis, the diagnostic results are presented according to the first classification indicating the state of fundus lesions. In the second image diagnosis, the diagnostic results are displayed based on the second category, which has a different number of categories than the first category. The first image diagnostic device sends the first diagnostic result based on the first classification from the first image diagnosis to the first information processing device. The second image diagnostic device sends the second diagnostic result based on the second classification from the second image diagnostic to the first information processing device. The first information processing device sends the first diagnostic result and the second diagnostic result to the image acquisition device. The image acquisition device displays the combined diagnostic result after integrating the first diagnostic result and the second diagnostic result.

9. An information processing device, characterized in that, Equipped with a processor and storage device, The storage device maintains the patient's examined eye image data, additional information about the examined eye image data, and the correspondence between the additional information and the artificial intelligence performing image diagnosis on the examined eye image data. Based on the corresponding information and additional information from the examined eye image data, the processor determines at least one of a first artificial intelligence and a second artificial intelligence. The first artificial intelligence is used to perform a first image diagnosis of the examined eye image data, and the second artificial intelligence performs a second image diagnosis of the examined eye image data that differs from the first image diagnosis. The processor sends a second transmission data, including the image data of the examined eye and the determination information of the artificial intelligence, to the image diagnostic device containing the determined artificial intelligence.

Images