Medical information processing device, medical information processing system, medical information processing method and program

Abstract

When lesion analysis processing is performed by computer, it mainly provides medical images suitable for the lesion analysis processing. The intermediary device includes an image acquisition unit that acquires medical images and a determination unit that determines whether to perform computer-based lesion detection processing on the medical images. The determination unit determines whether to perform lesion detection processing based on at least a condition of whether specific image processing has been performed, or whether to perform lesion detection processing using a trained model.

Description

[Technical Field]

[0001] The present invention relates to a medical information processing device, a medical information processing system, a medical information processing method, and a program. [Background technology]

[0002] There are AI analysis systems that use AI (artificial intelligence) to analyze medical images acquired by modalities such as radiological imaging devices, and provide doctors with the analysis results to support their diagnoses. AI analysis systems perform image analysis using medical images sent from a server, etc., but there are cases where the medical images contain images that are inappropriate for AI analysis. In such cases, there is a risk that inaccurate analysis results will be provided to doctors.

[0003] Patent Document 1 describes a medical image processing device that determines whether a medical image is an image to be processed. The medical image processing device determines whether the medical image is an image to be processed in a predetermined process based on the determination result of at least one of the imaging region and imaging direction of the medical image. Next, the medical image processing device performs rib attenuation processing, lesion detection processing, etc. on the medical image determined to be an image to be processed. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent No. 7035569 Summary of the Invention [Problem to be solved by the invention]

[0005] However, applying conventional technology to AI analysis poses the following challenges: Conventional medical image processing devices determine whether a medical image is a target image for a given process based on the results of determining the imaging location and imaging direction of the medical image. As a result, even medical images that are unsuitable for AI analysis, such as bone weakening processed images, may be determined to be appropriate if they meet certain conditions, such as the imaging location.

[0006] Therefore, in order to solve the above problem, the present invention aims to provide a medical information processing device, a medical information processing system, a medical information processing method, and a program that can provide only medical images suitable for lesion analysis processing when performing lesion analysis processing by computer. [Means for solving the problem]

[0007] The medical information processing device according to the present invention comprises: an image acquisition unit for acquiring medical images; a determination unit that determines whether or not to perform a lesion detection process by computer on the medical image, The determination unit determines whether or not to perform the lesion detection process based on at least a condition of whether or not specific image processing has been performed, or determines whether or not to perform the lesion detection process using a trained model.

[0008] The medical information processing system according to the present invention comprises: an image acquisition unit for acquiring medical images; a determination unit that determines whether or not to perform a lesion detection process by computer on the medical image, The determination unit determines whether or not to perform the lesion detection process based on at least a condition of whether or not specific image processing has been performed, or determines whether or not to perform the lesion detection process using a trained model.

[0009] The medical information processing method according to the present invention comprises: an image acquisition step of acquiring a medical image; a determination step of determining whether or not to perform lesion detection processing by a computer on the medical image, In the determination step, it is determined whether or not to perform the lesion detection process based on at least the condition of whether or not specific image processing has been performed, or it is determined whether or not to perform the lesion detection process using a trained model.

[0010] The program according to the present invention comprises: Computer, an image acquisition unit for acquiring medical images; a determining unit that determines whether or not to perform computer-based lesion detection processing on a medical image; The determination unit determines whether or not to perform the lesion detection process based on at least a condition of whether or not specific image processing has been performed, or determines whether or not to perform the lesion detection process using a trained model. [Effects of the Invention]

[0011] According to the present invention, whether or not to perform computer-based lesion detection processing on a medical image is determined, so that computer-based lesion detection processing can be performed only on medical images that are suitable for lesion detection processing. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a diagram illustrating an example of a schematic configuration of a medical information processing system according to a first embodiment. [Figure 2A] FIG. 2 is a diagram illustrating an example of a block diagram of an intermediary device according to the first embodiment. [Figure 2B] FIG. 2 is a diagram showing an example of a digital marker added to a medical image that is a radiological image according to the first embodiment. [Figure 3] FIG. 1 is a diagram showing an example of a block diagram of an image storage device according to a first embodiment. [Figure 4] 10 is a flowchart showing an example of the operation of the medical information processing system when determining whether or not to execute lesion detection processing on a medical image based on the rule base according to the first embodiment. [Figure 5A] FIG. 10 is a diagram showing an example of lesion analysis impossible information displayed on a display unit of a client terminal of the image storage device according to the first embodiment. [Figure 5B] FIG. 10 is a diagram showing another example of the lesion analysis impossible information displayed on the display unit of the client terminal of the image storage device according to the first embodiment. [Figure 6] 10 is a flowchart showing an example of the operation of the medical information processing system when determining whether or not to execute lesion detection processing using a trained model according to the first embodiment. [Figure 7] 10 is a flowchart showing an example of the operation of the medical information processing system in a case where a rule base according to the first embodiment and a determination as to whether to perform lesion detection processing using a trained model are combined. [Figure 8] 10 is a flowchart showing an example of the operation of the medical information processing system when performing lesion detection processing using a trained model according to the first embodiment. [Figure 9] FIG. 10 is a diagram illustrating an example of a schematic configuration of a medical information processing system according to a second embodiment. [Figure 10] FIG. 10 is a diagram illustrating an example of a schematic configuration of a medical information processing system according to a third embodiment. [Figure 11] FIG. 10 is a block diagram of an image capturing and storing device according to a third embodiment. [Figure 12] FIG. 10 is a block diagram of an image capturing and storing device according to a third embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0013] Preferred embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings.

[0014] [Configuration example of medical information processing system 100A] FIG. 1 is a diagram showing an example of the schematic configuration of a medical information processing system 100A according to the first embodiment. The medical information processing system 100A includes a hospital information system 1, a radiology information system 2, an imaging device 3, an intermediary device 4, and an image storage device 5. The hospital information system 1, the radiology information system 2, the imaging device 3, the intermediary device 4, and the image storage device 5 are communicably connected to each other via a network N1 established within a medical institution. An example of the network N1 is a LAN. LAN is an abbreviation for Local Area Network. The network N1 may be wired or wireless. Each system and device transmits and receives image data such as medical images to and from each other in accordance with the DICOM standard. DICOM is an abbreviation for Digital Imaging and Communication in Medicine.

[0015] The hospital information system 1 is a system that manages medical treatment, accounting, electronic medical records, etc. of the entire medical institution. The hospital information system 1 transmits test orders, such as tests and prescriptions performed by doctors, nurses, etc., to the imaging device 3, etc. This hospital information system 1 is called HIS, which is an abbreviation for Hospital Information System.

[0016] The radiology information system 2 is a system that manages imaging information such as examination order reservations, irradiation record management, and examination work statistics in a database, mainly used in the radiology department. The radiology information system 2 transmits examination orders, such as examinations and prescriptions performed by doctors, nurses, etc., to the imaging device 3, etc. The radiology information system 2 can be linked to the hospital information system 1. The hospital information system 1 and the radiology information system 2 may be integrated to form a single system. This radiology information system 2 is called RIS, which is an abbreviation for Radiology Information System.

[0017] The imaging device 3 images a subject based on an examination order received from the hospital information system 1 and the radiology information system 2, and generates a medical image of a predetermined imaging region, etc. The imaging device 3 is, for example, a CR device, an FPD device, a CT device, an MRI device, etc. The medical image is, for example, a radiological image. CR is an abbreviation for Computed Radiography. FPD is an abbreviation for Flat Panel Detector. CT is an abbreviation for Computed Tomography. MRI is an abbreviation for Magnetic Resonance Imaging. The imaging device 3 may perform specific image processing on the generated medical image. Examples of the specific image processing include bone weakening processing, as will be described later.

[0018] The imaging device 3 transmits the generated medical images to the intermediary device 4 in a file format conforming to the DICOM standard. Here, a DICOM standard file consists of an image section and a header section. The actual data of the medical image is written in the image section. The header section writes additional information related to the medical image. The additional information includes, for example, patient information, examination information, a UID for uniquely identifying the medical image, and information indicating whether specific image processing has been performed. The patient information includes patient identification information such as a patient ID for identifying the patient, as well as information such as the patient's name, gender, and date of birth. The examination information includes examination identification information such as an examination ID for identifying the examination, the examination date, the imaging area, the imaging direction, modality information, etc.

[0019] The intermediary device 4 is an example of a medical information processing device and is referred to as, for example, a gateway. The intermediary device 4 determines whether to perform computer-based lesion detection processing on medical images transmitted from the imaging device 3. Computer-based lesion detection processing is processing that determines whether a medical image is suitable for computer-based lesion detection processing using, for example, a trained model created by machine learning or deep learning, a type of machine learning. The intermediary device 4 determines whether to perform lesion detection processing based on at least a condition of whether specific image processing has been performed, or determines whether to perform lesion detection processing using a trained model. The intermediary device 4 performs computer-based lesion detection processing on the medical image and generates an analysis result. The intermediary device 4 transmits the generated analysis result along with the medical image to the image storage device 5. Note that in this embodiment, the determination of whether to perform computer-based lesion detection processing is performed by the intermediary device 4, but is not limited thereto. For example, the determination of whether to perform computer-based lesion detection processing may be performed by the imaging device 3, the image storage device 5, etc., or by an image imaging and storage device that integrates the imaging device 3 and the image storage device 5. Furthermore, the decision as to whether or not to perform lesion detection processing by computer may be made by an image analysis device 9 (see FIG. 9) or the like provided on the cloud.

[0020] The image storage device 5 stores and manages the image data of medical images and their accompanying information sent from the intermediary device 4. The image storage device 5 stores and manages the analysis results of the lesion analysis process sent from the intermediary device 4. The image storage device 5 reads out the corresponding medical image in response to a search request for a specific medical image from a user such as a doctor, and transmits the read medical image to the client terminal, viewer, etc. that made the request.

[0021] In the above medical information processing system 100A, the imaging device 3 transmits medical images to the intermediary device 4, and the intermediary device 4 transmits the medical images and their analysis results to the image storage device 5, but this is not limiting. For example, the imaging device 3 may transmit medical images to both the intermediary device 4 and the image storage device 5, and the intermediary device 4 may transmit only the analysis results of the medical images to the image storage device 5.

[0022] [Configuration example of intermediary device 4] Next, the configuration of the intermediary device 4 according to the first embodiment will be described. Fig. 2A is a block diagram of the intermediary device 4 according to the first embodiment. The intermediary device 4 includes a control unit 40, an operation unit 41, a display unit 42, a storage unit 43, and a communication unit 44. The control unit 40, the operation unit 41, the display unit 42, the storage unit 43, and the communication unit 44 are connected to each other via wiring such as a bus 47.

[0023] The control unit 40 includes, for example, a processor such as a CPU that performs calculations and control, and memory. CPU is an abbreviation for Central Processing Unit. The control unit 40 executes a program stored in, for example, the storage unit 43, thereby performing processes such as determining whether or not to perform computer-based lesion detection processing on medical images. The control unit 40 may also include electronic circuits such as ASIC and FPGA. ASIC is an abbreviation for Application Specific Integrated Circuit. FPGA is an abbreviation for Field Programmable Gate Array.

[0024] In this embodiment, the control unit 40 functions as an acquisition unit 400, a determination unit 401, an analysis unit 402, and an output unit 403. A processor or the like of the control unit 40 executes a program stored in a storage unit 43 or the like to realize the functions of the acquisition unit 400, the determination unit 401, the analysis unit 402, and the output unit 403. Details of each function of the control unit 40 will be described later.

[0025] The operation unit 41 includes, for example, a mouse, keyboard, switches, buttons, etc. The operation unit 41 may be, for example, a touch panel integrally combined with a display, or an interface that accepts voice input. The operation unit 41 accepts instructions corresponding to various input operations from the user, converts the accepted instructions into operation signals, and outputs them to the control unit 40. Specifically, the operation unit 41 accepts instructions such as a selection instruction as to whether to perform computer-based lesion detection processing on medical images that may contain foreign objects.

[0026] The display unit 42 is, for example, a liquid crystal display, an organic EL display, or the like. EL is an abbreviation for Electro Luminescence. The display unit 42 displays an image that has undergone predetermined analysis processing, a GUI for accepting various input operations from the user, and the like. GUI is an abbreviation for Graphical User Interface. Specifically, the display unit 42 displays a message or the like to notify the user that a foreign object may be captured in the medical image.

[0027] The storage unit 43 includes any storage module, such as an HDD, SSD, ROM, and RAM. HDD is an abbreviation for Hard Disk Drive. SSD is an abbreviation for Solid State Drive. ROM is an abbreviation for Read Only Memory. The storage unit 43 stores, for example, system programs, application programs, and various data. Specifically, the storage unit 43 stores a program that determines whether or not to perform computer-based lesion detection processing on medical images, and image data of medical images captured by the imaging device 3. The storage unit 43 also stores a trained model that outputs a determination result as to whether or not to perform computer-based lesion detection processing on medical images.

[0028] The communication unit 44 includes, for example, a communication module including a NIC, a receiver, and a transmitter. NIC is an abbreviation for Network Interface Card. The communication unit 44 communicates various information and image data with the imaging device 3, the image storage device 5, etc. via the network N1.

[0029] 2A. For example, the intermediary device 4 may be configured without the operation unit 41 and the display unit 42. In this case, the intermediary device 4 automatically processes the medical images sent from the imaging device 3 without receiving instructions from the user, and sends the analysis results of the processed medical images to the image storage device 5.

[0030] [Example of function of intermediary device 4] Next, the functions of the acquisition unit 400, determination unit 401, analysis unit 402, and output unit 403 of the control unit 40 of the intermediary device 4 will be described in detail. In this embodiment, the control unit 40 determines whether or not to perform computer-based lesion detection processing on a medical image. Specifically, the determination unit 401 determines whether or not to perform lesion detection processing based on at least a condition of whether specific image processing has been performed, or determines whether or not to perform lesion detection processing using a trained model.

[0031] The communication unit 44 functions as an image acquisition unit and acquires medical images captured by the imaging device 3. The acquisition unit 400 acquires information such as whether specific image processing has been performed from the additional information of the medical image or by image recognition of the medical image. The additional information of the medical image is tag information of the medical image, information managed in a database separately from the medical image and stored in association with the medical image.

[0032] Specifically, the acquiring unit 400 acquires at least one of the following information from the tag information of the medical image: information indicating whether specific image processing has been performed, whether the imaging conditions are standard, whether the medical image satisfies specific image quality conditions, and whether the medical image is an image captured by a modality other than a specific modality.

[0033] The acquisition unit 400 acquires at least one of the following information by image recognition of the medical image. The information includes information indicating whether specific image processing has been performed, whether the imaging conditions are standard, whether the medical image satisfies specific image quality conditions, and whether the detection target site to be detected in the lesion detection processing is missing. Other information includes information indicating whether the medical image contains a foreign object that is not a detection target, and whether the medical image is an image captured by a modality other than the specific modality.

[0034] Among the above-mentioned information, information indicating whether a detection target portion to be detected in the lesion detection process is missing or whether a foreign object that is not a detection target is included in the medical image can basically only be obtained by image recognition. Whether a detection target portion to be detected in the lesion detection process is missing or not may be determined by whether there is a missing lung field in the chest lesion detection process. Whether a foreign object that is not a detection target is included in the medical image may be determined by whether a low signal value region that is suspected to be a foreign object is included by histogram analysis.

[0035] The acquisition unit 400 may acquire information such as whether specific image processing has been performed from information contained in the hospital information system 1 or information contained in the radiology information system 2. Hereinafter, the information contained in the hospital information system 1 will be referred to as HIS information, and the information contained in the radiology information system 2 will be referred to as RIS information. The supplementary information of a medical image may not contain sufficient information to determine whether or not to perform lesion detection processing, and image recognition of the medical image may be time-consuming. By using HIS information or RIS information, it is possible to improve the accuracy of acquiring information indicating whether or not a specific image processing has been performed on a medical image.

[0036] The HIS information includes imaging instruction information, examination orders, etc. For example, information indicating whether or not a patient uses a pacemaker can be obtained from the HIS information. If a radiology information system 2 is interposed between the hospital information system 1 and the imaging device 3, etc., the HIS information may be acquired via the radiology information system 2. Information regarding images that are not subject to restrictions, etc., is also stored in the HIS information or RIS information.

[0037] The determination unit 401 determines whether specific image processing has been performed based on a condition. Specifically, the determination unit 401 uses information on whether specific image processing has been performed acquired by the acquisition unit 400 to determine whether the specific image processing is appropriate for lesion detection processing based on a predetermined condition. That is, in this determination method, a rule-based determination is made as to whether lesion detection processing by a computer should be performed.

[0038] The specific image processing is at least one process selected from the group of processes listed below. Examples of the group of processes include vertical and horizontal inversion, rotation, black and white inversion, For Processing, bone attenuation processing, and temporal subtraction processing. Other group of processes include energy subtraction, outside-irradiation-field blackening, frequency processing above a certain intensity, addition of digital markers, and addition of annotations. For example, frequency processing above a certain intensity may emphasize human body structures and result in them being mistakenly recognized as lesions. Digital markers and annotations may be considered foreign bodies and affect the accuracy of lesion detection processing. Figure 2B shows an example of a digital marker M added to a medical image G, which is a radiographic image according to the first embodiment. A medical image G of the front of the chest acquired by the acquisition unit 400 includes a digital marker M added to indicate that the imaging direction of the medical image G is the PA direction. The digital marker M may be considered a foreign body in computer-based lesion detection processing. A blackening-processed image is, for example, an image in which blackening processing with a width of 40 mm or more is applied to the top edge of the image. An example of a bone attenuation processed image is a clavicle / rib attenuation processed image.

[0039] The determination unit 401 can determine whether or not to perform lesion detection processing based on conditions such as whether or not specific image processing has been performed, using information acquired from the tag information of the medical image. Specifically, the determination unit 401 determines whether or not specific image processing has been performed by checking whether or not the value of "(0008,0008)_Image_Type" defined by the DICOM standard is "ORIGINAL", which indicates that the image is an original image. The determination unit 401 also determines whether or not specific image processing has been performed by checking whether or not the value of "(0008,2111)_Derivation Description" is "BONE SUPPRESSION". The judgment is made.

[0040] The determination unit 401 can determine whether to perform computer-based lesion detection processing on a medical image based on at least one of the following conditions. Examples of the conditions include whether the medical image is captured under standard imaging conditions, whether the medical image satisfies specific image quality conditions, and whether the target region to be detected in the lesion detection processing is missing. Other conditions include whether the medical image contains a foreign object that is not a detection target, and whether the medical image was captured using a modality other than a specific modality. Another condition is whether the image was captured under specific imaging conditions other than the imaging direction or imaging region. Still other conditions include whether the imaging direction is PA direction in the case of chest imaging in an upright position, whether the imaging direction is AP direction in the case of imaging in a supine or sitting position, and whether the imaging direction is CC direction in the case of breast imaging. These conditions may be added before or after the determination of whether to perform lesion detection processing based on whether specific image processing has been performed, or the determination of whether to perform lesion detection processing using a trained model.

[0041] Whether or not the medical image satisfies the standard imaging conditions is determined by whether or not the tilt of the medical image satisfies the conditions, whether or not the medical image is a portable imaging image, whether or not the medical image is an enlarged imaging image, or whether or not the medical image is a spot imaging image.

[0042] The tilt of a medical image satisfies the condition, for example, when the tilt of the center line of the lung field is ±10 degrees or more from the vertical as a result of lung field recognition processing of the image. Whether a medical image is a portable image can be determined, for example, by deep learning. Whether a medical image is a magnified image is determined by whether the value of "(0054,0222)_View Modifier Code Sequence_View Modifier Code Sequence" defined by the DICOM standard in the tag information is [Magnification]. Whether a medical image is a magnified image is determined by whether the value of "(0018,1114)_Estimated Radiographic Magnification Factor_Estimated Radiographic Magnification Factor" defined by the DICOM standard in the tag information is other than 1. Whether a medical image is a spot image is determined by whether the value of "(0054,0222)_View Modifier Code Sequence_View Modifier Code Sequence" defined by the DICOM standard in the tag information is "Spot Compression." Furthermore, whether or not a medical image is a spot image is determined when the "(0018,1114)_Estimated Radiographic Magnification Factor_Estimated Radiographic Magnification Factor" defined by the DICOM standard in the tag information is a value other than 1.

[0043] Furthermore, whether or not the medical image is under standard imaging conditions refers to whether or not the imaging position of the subject in the medical image satisfies the conditions, whether or not the dose information of the medical image satisfies the conditions, and whether or not the medical image is an image in which moire occurs. The tilt of the medical image includes, for example, the tilt of the lung field of the subject shown in the medical image. The determination based on the dose information of the medical image is, for example, whether or not the dose value or the dose estimate value is equal to or greater than a threshold.

[0044] The imaging position of the subject in a medical image meets the conditions if the tag information defined by the DICOM standard does not contain the word "standing." The dose information of a medical image meets the conditions if, in the case of a frontal chest image, the dose information is less than 0.5 mAs. Moiré in a medical image can be determined by power spectrum analysis if a peak is detected at a specific frequency, for example, a peak 50 times or more the average signal value at 2.8 cycles / mm.

[0045] Whether a medical image satisfies a specific image quality condition is determined by, for example, whether the pixel spacing is 200 μm or less in both the vertical and horizontal directions. If the "(0028,2112) Lossy Image Compression Ratio_lossy image compression ratio" defined by the DICOM standard is 10 or greater, the determination unit 401 determines that the medical image is unsuitable for computer-based lesion detection processing.

[0046] Whether a medical image satisfies specific image quality conditions depends on at least one of whether the medical image is a film digitized image, whether the resolution of the medical image satisfies the conditions, and whether lossy compression processing has been performed.

[0047] Whether or not the detection target area to be detected in the lesion detection process is missing is determined by at least one of whether or not the medical image is a lung field missing image and whether or not the imaging target area is cut off outside the medical image.

[0048] Whether or not a medical image contains a foreign object that is not a detection target is determined by at least one of whether or not the medical image contains a foreign object other than a living body and whether or not a large lesion is present in the medical image. The foreign object other than a living body may be at least one selected from, for example, a pacemaker, wires, a spinal fixation corset, postoperative scars, post-pectomy chest surgery, a wristwatch, metal underwear, piercings, markers, compresses, and silicone. Markers include markers in the lung field and metal markers placed in the breast.

[0049] Whether a medical image is an image captured by a modality other than the specific modality is determined based on whether the medical image is at least one of a tomographic image, an ultrasound image, and a video image. Specifically, whether a medical image is an image captured by a modality other than the specific modality is determined based on whether the medical image is an image captured by a modality different from the modality that captured the medical image used as learning data for the lesion detection process by the computer.

[0050] Determination unit 401 may determine whether to perform computer-based lesion detection processing depending on the type of lesion detection processing. The type of lesion detection processing is determined by the subject, the imaging region, the type of imaging device, etc. For example, the types of lesion detection processing include AI lesion detection processing for chest images, which are still X-ray images, AI lesion detection processing for breast images, etc.

[0051] Furthermore, the determination unit 401 can directly determine whether or not to perform lesion detection processing using a trained model. Specifically, the trained model is a model trained by machine learning using a dataset in which images appropriate or inappropriate for lesion detection processing are paired with correct labels indicating whether the images are appropriate or inappropriate for lesion detection processing. Preferably, the trained model is a model trained by machine learning using a dataset in which at least medical images that have undergone specific image processing or medical images that have not undergone image processing are paired with correct labels indicating whether the images are appropriate or inappropriate for lesion detection processing. The determination unit 401 inputs a medical image to the trained model and determines whether or not to perform lesion detection processing by obtaining, as an output result from the trained model, information indicating whether the medical image is appropriate or inappropriate for lesion detection processing.

[0052] The determination unit 401 can determine whether to perform lesion detection processing based on a trained model that uses medical images containing at least one of the following information as training data. Examples of the information include whether the imaging conditions are standard, whether the medical image satisfies specific image quality conditions, and whether the target area to be detected in the lesion detection processing is missing. Other examples of the information include whether the medical image contains a foreign object that is not a detection target, and whether the medical image was captured using a modality other than a specific modality.

[0053] For example, if the medical image is suitable for the lesion detection process, the analysis unit 402 performs the lesion detection process by analyzing the medical image using a second trained model trained by machine learning. The second trained model is a model trained by machine learning using, for example, multiple data sets of medical images including lesions and medical images not including lesions, and analysis results based on the lesions.

[0054] The output unit 403 outputs a medical image based on the determination of whether or not to perform computer-based lesion detection processing on the medical image. When the determination unit 401 determines that computer-based lesion detection processing should not be performed on the medical image, the output unit 403 outputs information indicating the reason, basis, etc. for the determination result to the display unit 56 of the image storage device 5. When making a diagnosis, a user such as a doctor opens a screen on the display unit 56 on which the medical image can be referenced. Therefore, by displaying the reason, basis, etc. for the determination result of whether or not to perform computer-based lesion detection processing as an image on the same screen as the medical image, diagnostic work can be performed efficiently without overlooking anything.

[0055] [Configuration example of Image Archiving Device 5] Next, the configuration of the image storage device 5 according to the first embodiment will be described. Fig. 3 is a block diagram of the image storage device 5 according to the first embodiment. Note that the description of components having the same configuration and functions as the intermediary device 4 described in Fig. 2A will be omitted or simplified.

[0056] The image storage device 5 includes a server 5a and a client terminal 5b. The server 5a and the client terminal 5b are connected via a LAN or the like. The server 5a includes a control unit 50, an operation unit 51, a storage unit 53, and a communication unit 54. The control unit 50, the operation unit 51, the storage unit 53, and the communication unit 54 are connected to one another via wiring such as a bus 57.

[0057] The control unit 50 includes, for example, a processor such as a CPU that performs calculations and control, a memory, etc. The control unit 50 executes a program stored in, for example, the storage unit 53, etc., thereby realizing processes such as storing and managing medical images, etc.

[0058] In this embodiment, the control unit 50 functions as an acquisition unit 500 and an output unit 503. A processor or the like of the control unit 50 executes a program stored in a storage unit 53 or the like to realize the functions of the acquisition unit 500 and the output unit 503. The acquisition unit 500 acquires, for example, medical images to which analysis results of a computer-based lesion detection process have been added, transmitted from the intermediary device 4. When a display request for a predetermined medical image is received from a user such as a doctor, the output unit 503 outputs the corresponding medical image to the client terminal 5b.

[0059] The operation unit 51 includes, for example, a mouse, a keyboard, switches, buttons, etc. The operation unit 51 receives instructions in response to various input operations from the user, converts the received instructions into operation signals, and outputs the operation signals to the control unit 50.

[0060] The storage unit 53 includes any storage module, such as an HDD, SSD, ROM, and RAM. For example, system programs, application programs, and various data are stored in the storage unit 53. Specifically, the storage unit 53 stores image data such as medical images to which the analysis results of computer-based lesion detection processing have been added, and medical images to which computer-based lesion detection processing has not been performed.

[0061] The communication unit 54 includes, for example, a communication module including a NIC, a receiver, and a transmitter. The communication unit 54 communicates various information and image data with the intermediary device, etc. via the network N1. The communication unit 54 also communicates image data, etc. with the client terminal 5b.

[0062] The display unit 56 of the client terminal 5b is, for example, a liquid crystal display, an organic EL display, etc. The display unit 56 displays medical images to which the analysis results of the lesion detection processing by a computer have been added, medical images to which the lesion detection processing by a computer has not been performed, etc.

[0063] [Example of operation of medical information processing system 100A] (1. Method for determining whether or not to perform rule-based lesion detection processing) Next, a description will be given of the flow of the first medical information processing method in the medical information processing system 100A according to the first embodiment. Fig. 4 is a flowchart showing an example of the operation of the medical information processing system 100A when determining whether or not to execute lesion detection processing on a medical image based on the rule base according to the first embodiment.

[0064] The communication unit 44 of the intermediary device 4 acquires the medical images captured by the imaging device 3 (step S10). Step S10 corresponds to an image acquisition step. The medical images acquired by the communication unit 44 are stored in the storage unit 43.

[0065] The acquiring unit 400 acquires tag information attached to the acquired medical image from the medical image (step S11). The tag information is defined, for example, by the DICOM standard. When specific image processing such as rib attenuation processing is performed on the medical image, processing information indicating that the specific image processing has been performed is added to the tag information.

[0066] The determination unit 401 of the intermediary device 4 determines whether the content of the tag information acquired by the acquisition unit 400 is information indicating a medical image suitable for computer-based lesion detection processing (step S12). Step S12 corresponds to the determination step. Specifically, the determination unit 401 determines whether the value of "(0008,2111)_Derivation Description_Derivation Description" defined by the DICOM standard is "BONE SUPPRESSION". "BONE SUPPRESSION" indicates that rib attenuation processing has been performed on the medical image. Furthermore, the determination unit 401 determines whether the value of "(0008,0008)_Image Type" defined by the DICOM standard is "DERIVED\PRIMARY". "DERIVED\PRIMARY" indicates that image processing that is significantly different from the original image has been performed. Indicates a possible derivative image.

[0067] If the content of the tag information is not "BONE SUPPRESSION" or "DERIVED\PRIMARY", the determination unit 401 proceeds to step S13. In other words, the content of the tag information indicates that the medical image is suitable for lesion detection processing by a computer. In this case, the determination unit 401 determines that lesion detection processing by a computer is to be performed on the acquired medical image (step S13). Step S13 corresponds to a determination step.

[0068] The analysis unit 402 of the intermediary device 4 performs a computer-based lesion detection process on the medical image (step S14). Specifically, the analysis unit 402 inputs the acquired medical image into a trained model and acquires a medical image with the lesion analysis result attached as an output result from the trained model. The output unit 403 outputs the medical image with the lesion analysis result attached to the communication unit 44. The communication unit 44 transmits the medical image to the image storage device 5 via the network N1.

[0069] The acquisition unit 500 of the client terminal 5b of the image storage device 5 acquires medical images, etc. from the intermediary device 4 via the communication unit 54 and stores the acquired medical images, etc. in the storage unit 53. When a search instruction for a specific medical image is received from a user such as a doctor via the operation unit 51, the acquisition unit 500 acquires the corresponding medical image from the storage unit 53. The output unit 503 outputs the acquired medical image to the client terminal 5b. The display unit 56 of the client terminal 5b displays on the screen the specific medical image searched for by the user and the lesion analysis results assigned to the medical image (step S15). This allows a user such as a doctor to quickly and accurately diagnose a patient by checking the lesion analysis results displayed on the screen.

[0070] On the other hand, in step S12, if the content of the tag information is "BONE SUPPRESSION" or "DERIVED\PRIMARY," the determination unit 401 of the intermediary device 4 proceeds to step S16. In other words, the content of the tag information indicates that the medical image is not suitable for computer-based lesion detection processing. In this case, the determination unit 401 determines that the acquired medical image is not suitable for computer-based lesion detection processing (step S16). For example, the determination unit 401 adds lesion analysis impossibility information to the medical image, indicating the reason and basis why computer-based lesion detection processing is not possible with the currently acquired medical image. The output unit 403 outputs the medical image and the lesion analysis impossibility information to the communication unit 44. The communication unit 44 transmits the medical image and the lesion analysis impossibility information to the image storage device 5 via the network N1.

[0071] The acquisition unit 500 of the client terminal 5b of the image storage device 5 acquires the medical images, etc. and the lesion analysis impossible information from the intermediary device 4 via the communication unit 54, and stores the acquired medical images, etc. in the storage unit 53. When a search instruction for a specific medical image is received from a user via the operation unit 51, the acquisition unit 500 acquires the medical image from the storage unit 53. The output unit 503 outputs the acquired medical image and the lesion analysis impossible information to the client terminal 5b. The display unit 56 of the client terminal 5b displays the specific medical image searched for by the user and the lesion analysis impossible information on the screen (step S17).

[0072] 5A is a diagram showing an example of the lesion analysis impossibility information I displayed on the display unit 56 of the client terminal 5b of the image storage device 5 according to the first embodiment. The lesion analysis impossibility information I, which indicates the reason and basis for why lesion detection processing by computer cannot be performed on the current medical image, is displayed on the screen of the display unit 56. For example, if rib attenuation processing has already been performed on the medical image, the lesion analysis impossibility information I displayed on the screen of the display unit 56 reads, "Lesion detection cannot be performed on medical images that have been subjected to BS (rib attenuation) processing."

[0073] 5B is a diagram showing another example of the lesion analysis impossibility information I displayed on the display unit 56 of the client terminal 5b of the image storage device 5 according to the first embodiment. The lesion analysis impossibility information I is displayed on the screen of the display unit 56 in addition to the medical image G. For example, the lesion analysis impossibility information I is displayed on the screen of the display unit 56 in the form of a message superimposed on the current medical image G, stating that "lesion detection cannot be performed on medical images that have undergone BS processing."

[0074] (2. How to determine whether to perform lesion detection processing using a trained model) Next, the flow of the second medical information processing method in the medical information processing system 100A according to the first embodiment will be described. Fig. 6 is a flowchart showing an example of the operation of the medical information processing system 100A when determining whether or not to execute lesion detection processing using a trained model according to the first embodiment. Note that detailed explanations of processes common to the first medical information processing method in Fig. 4 will be omitted or simplified.

[0075] The communication unit 44 of the intermediary device 4 acquires the medical images captured by the imaging device 3 (step S20). The medical images acquired by the communication unit 44 are stored in the storage unit 43.

[0076] The determination unit 401 of the intermediary device 4 executes a lung field defect detection process for the acquired medical image (step S21). For example, the determination unit 401 may execute the lung field defect detection process for the medical image using a process based on machine learning. For example, the lung field defect detection process may employ the technology described in Japanese Patent Application Laid-Open No. 2013-102848. Specifically, the determination unit 401 calculates a predetermined first feature value for a partial region within the irradiation field in the medical image that is adjacent to the boundary of the irradiation field. The determination unit 401 calculates a predetermined second feature value that indicates the presence or absence of a high-density region outside the lung field from the medical image. Next, the determination unit 401 inputs the calculated first feature value and second feature value into a trained model, and obtains defect information regarding the presence or absence of a defect in the lung field of the medical image as an output result from the trained model.

[0077] Here, the trained model can be generated by training a computer with machine learning capabilities using predetermined learning data. Specifically, the computer prepares multiple sample images, including images with and without lung field defects, and calculates first and second feature amounts from each sample image. The computer performs machine learning to determine, based on the distribution of the first and second feature amounts, what boundary line should be drawn to distinguish between images with and without lung field defects. The computer outputs defect information indicating whether the image is defect-free or defect-containing, depending on which side of the boundary line the input feature amount is located on. The trained model is stored, for example, in the storage unit 43 of the intermediary device 4. The machine learning algorithm is not particularly limited, but examples that can be used include Adaboost, support vector machines, and neural networks.

[0078] The determination unit 401 of the intermediary device 4 determines whether or not the acquired medical image has a lung field defect (step S22). If the defect information output from the trained model indicates "no lung field defect," the determination unit 401 determines that the medical image has no lung field defect and proceeds to step S23. In this case, the determination unit 401 determines that the acquired medical image is to undergo computer-based lesion detection processing (step S23).

[0079] The analysis unit 402 of the intermediary device 4 performs a computer-based lesion detection process on the medical image (step S24). Specifically, the analysis unit 402 inputs the acquired medical image into a trained model and acquires a medical image with an analysis result of the lesion attached as an output result from the trained model. The communication unit 44 transmits the medical image output from the output unit 403 to the image storage device 5 via the network N1.

[0080] The acquisition unit 500 of the client terminal 5b of the image storage device 5 acquires medical images, etc. from the intermediary device 4 via the communication unit 54, and stores the acquired medical images, etc. in the storage unit 53. When a search instruction for a specific medical image is received from a user via the operation unit 51, the acquisition unit 500 acquires the medical image from the storage unit 53. The display unit 56 of the client terminal 5b displays on the screen the specific medical image searched for by the user and the analysis result of the lesion assigned to the medical image (step S25).

[0081] On the other hand, in step S22, if the defect information output from the trained model indicates "lung field defect," the determination unit 401 determines that the medical image has a lung field defect and proceeds to step S26. In this case, the determination unit 401 determines that the acquired medical image is not suitable for computer-based lesion detection processing (step S26). For example, the determination unit 401 adds lesion analysis impossibility information indicating the reason and basis why computer-based lesion detection processing cannot be performed on the currently acquired medical image to the medical image. The communication unit 44 transmits the medical image and the lesion analysis impossibility information output from the output unit 403 to the image storage device 5 via the network N1.

[0082] The acquisition unit 500 of the client terminal 5b of the image storage device 5 acquires the medical images and the lesion analysis impossible information from the intermediary device 4 via the communication unit 54, and stores the acquired medical images, etc. in the storage unit 53. When a search instruction for a specific medical image is received from the user via the operation unit 51, the acquisition unit 500 acquires the medical image from the storage unit 53. The display unit 56 of the client terminal 5b displays the specific medical image searched for by the user and the lesion analysis impossible information on the screen (step S27).

[0083] (3. Method for determining whether to perform lesion detection processing using rule-based and trained models) Next, the flow of the third medical information processing method in the medical information processing system 100A according to the first embodiment will be described. Fig. 7 is a flowchart showing an example of the operation of the medical information processing system 100A when the rule base according to the first embodiment and the determination of whether to perform lesion detection processing using a trained model are combined. Note that the description of processes common to the first medical information processing method in Fig. 4 will be simplified or omitted.

[0084] The communication unit 44 of the intermediary device 4 acquires the medical image captured by the imaging device 3 (step S30). The acquisition unit 400 acquires DICOM standard tag information attached to the medical image from the acquired medical image (step S31).

[0085] The determination unit 401 of the intermediary device 4 determines whether the value of "(0008,2111)_Derivation Description_Derivation Description" defined by the DICOM standard in the tag information is "BONE SUPPRESSION." Furthermore, the determination unit 401 determines whether the value of "(0008,0008)_Image Type" defined by the DICOM standard in the tag information is "DERIVED\PRIMARY" (step S32). If the determination unit 401 determines that the content of the tag information is not "BONE SUPPRESSION" or "DERIVED\PRIMARY," the process proceeds to step S33.

[0086] The determination unit 401 acquires processing information indicating whether rib attenuation processing has been performed on the medical image acquired using the trained model (step S33). That is, in step S33, a secondary determination is made as to whether rib attenuation processing has been performed on the medical image. This is because the content of tag information may be erroneously input. For example, the trained model is a model trained by machine learning using a dataset in which an image that has undergone rib attenuation processing is paired with processing information indicating that the image is unsuitable for lesion detection processing. Alternatively, the trained model is a model trained by machine learning using a dataset in which an image that has not undergone rib attenuation processing is paired with processing information indicating that the image is suitable for lesion detection processing.

[0087] The determination unit 401 determines whether rib attenuation processing has been performed on the medical image based on the output result of the trained model (step S34). If the determination unit 401 determines that rib attenuation processing has not been performed on the medical image, the process proceeds to step S34. In this case, the determination unit 401 determines that computer-based lesion detection processing should be performed on the acquired medical image (step S35).

[0088] The determination unit 401 performs a computer-based lesion detection process on the acquired medical image (step S36). For example, the analysis unit 402 inputs the acquired medical image into a trained model and acquires a medical image with a lesion analysis result attached as an output result from the trained model. The output unit 403 outputs the medical image with the lesion analysis result attached to the communication unit 44. The communication unit 44 transmits the medical image to the image storage device 5 via the network N1.

[0089] The acquisition unit 500 of the client terminal 5b of the image storage device 5 acquires medical images from the intermediary device 4 via the communication unit 54. When a search instruction for a specific medical image is received from a user such as a doctor via the operation unit 51, the acquisition unit 500 acquires the medical image from the storage unit 53. The output unit 503 outputs the acquired medical image to the client terminal 5b. The display unit 56 of the client terminal 5b displays on the screen the specific medical image searched for by the user and the lesion analysis results assigned to the medical image (step S37). This allows a user such as a doctor to quickly and accurately diagnose a patient by checking the lesion analysis results displayed on the screen.

[0090] On the other hand, in step S32, if the content of the tag information is "BONE SUPPRESSION" or "DERIVED￥PRIMARY", the determination unit 401 proceeds to step S38. Also, in step S34, if the determination unit 401 determines from the output result of the trained learning model that rib attenuation processing is to be performed on the medical image, the determination unit 401 proceeds to step S38. In other words, this is the case when the acquired medical image is not suitable for lesion detection processing by a computer.

[0091] The determination unit 401 determines that computer-based lesion detection processing will not be performed on the acquired medical image (step S38). In this case, the determination unit 401 adds lesion analysis impossible information to the medical image, indicating that computer-based lesion detection processing cannot be performed on the currently acquired medical image. The output unit 403 outputs the medical image and the lesion analysis impossible information to the communication unit 44. The communication unit 44 transmits the medical image and the lesion analysis impossible information to the image storage device 5 via the network N1.

[0092] The acquisition unit 500 of the client terminal 5b of the image storage device 5 acquires the medical image and the lesion analysis impossible information from the intermediary device 4 via the communication unit 54. When a search instruction for a specific medical image is received from the user via the operation unit 51, the acquisition unit 500 acquires the medical image from the storage unit 53. The output unit 503 outputs the acquired medical image and the lesion analysis impossible information to the client terminal 5b. The display unit 56 of the client terminal 5b displays the specific medical image searched for by the user and the lesion analysis impossible information on the screen (step S39).

[0093] (4. Method of determining whether to perform lesion detection processing based on whether or not a foreign object is detected) Next, the flow of the fourth medical information processing method in the medical information processing system 100A according to the first embodiment will be described. Fig. 8 is a flowchart showing an example of the operation of the medical information processing system 100A when performing lesion detection processing using a trained model according to the first embodiment. Note that explanations of processes common to the first medical information processing method in Fig. 4 will be simplified or omitted.

[0094] The communication unit 44 acquires the medical images captured by the imaging device 3 (step S40). The medical images acquired by the communication unit 44 are stored in the storage unit 43.

[0095] The determination unit 401 executes a foreign body detection process to detect whether a foreign body is captured in the medical image (step S41). This is because if a computer-based lesion detection process is performed using a medical image that contains a foreign body, the accuracy of lesion detection may be significantly reduced due to the influence of the foreign body.

[0096] The determination unit 401 inputs the acquired medical image into a trained model, and acquires foreign object information indicating whether a foreign object is likely to be captured in the medical image as an output result from the trained model. The trained model is a model trained by machine learning using a dataset of medical images that capture foreign objects and foreign object information "captured" indicating that a foreign object is captured in the medical image. The trained model is also a model trained by machine learning using a dataset of medical images that do not capture foreign objects and foreign object information "not captured" indicating that a foreign object is not captured in the medical image.

[0097] If the acquired foreign substance information is "no foreign substance", the determination unit 401 determines that there is a low possibility that a foreign substance is captured in the medical image, and proceeds to step S43. In this case, the determination unit 401 determines that a computer-based lesion detection process is to be performed on the medical image (step S43).

[0098] The determination unit 401 performs a computer-based lesion detection process on the acquired medical image (step S44). For example, the analysis unit 402 inputs the acquired medical image into a trained model and acquires a medical image with lesion analysis results added as an output result from the trained model. The output unit 403 transmits the medical image with the lesion analysis results added to the image storage device 5 via the communication unit 44.

[0099] The acquisition unit 500 of the client terminal 5b of the image storage device 5 acquires medical images from the intermediary device 4 via the communication unit 54. When a search instruction for a specific medical image is received from a user such as a doctor via the operation unit 51, the acquisition unit 500 acquires the medical image from the storage unit 53. The output unit 503 outputs the acquired medical image to the client terminal 5b. The display unit 56 of the client terminal 5b displays on the screen the specific medical image searched for by the user and the lesion analysis results assigned to the medical image (step S45). This allows a user such as a doctor to quickly and accurately diagnose a patient by checking the lesion analysis results displayed on the screen.

[0100] On the other hand, in step S42, if the acquired foreign substance information is "reflected," the determination unit 401 determines that there is a high possibility that a foreign substance is reflected in the medical image, and proceeds to step S46. Here, if the foreign substance information is "reflected," certainty information indicating the certainty that a foreign substance is reflected in the medical image is further added to the foreign substance information. Therefore, based on the certainty information, the determination unit 401 determines whether the certainty that a foreign substance is reflected in the medical image is 80% or more (step S46).

[0101] If the confidence level is not 80% or higher, the determination unit 401 determines that what is shown in the medical image may not be a foreign object, and proceeds to step S47. The output unit 403 outputs confirmation information including a message to prompt a user, such as a doctor, to the display unit 42 to confirm whether or not a foreign object is shown in the medical image (step S47). The display unit 42 displays a message such as, "It is suspected that a foreign object is shown. Do you want to stop the lesion detection process as there is a possibility that lesion detection will not be performed correctly? If you want to continue the process, please confirm that no foreign object is shown and press [No]."

[0102] The determination unit 401 determines whether an instruction to stop the lesion detection process on the medical image has been acquired (step S48). If the user selects the "No" button from the operation unit 51, indicating that the lesion detection process should not be stopped, the determination unit 401 proceeds to step S44. In other words, this is the case when the user checks the medical image and determines that no foreign object is captured. In this case, as described above, the computer performs lesion detection process and the like on the acquired medical image.

[0103] On the other hand, in step S46, if the confidence level is 80% or higher, the determination unit 401 determines that the object in the medical image is a foreign object, and proceeds to step S49. Also, if the user selects the [Yes] button from the operation unit 51 to stop the lesion detection process, the determination unit 401 proceeds to step S49. In other words, this is the case when the user checks the medical image and determines that a foreign object is present.

[0104] The determination unit 401 determines that the acquired medical image is not suitable for computer-based lesion detection processing (step S49). In this case, the determination unit 401 adds lesion analysis impossibility information indicating the reason and basis why computer-based lesion detection processing is not possible with the currently acquired medical image to the medical image. The output unit 403 transmits the medical image and the lesion analysis impossibility information to the image storage device 5 via the communication unit 44.

[0105] The acquisition unit 500 of the client terminal 5b of the image storage device 5 acquires the medical image and the lesion analysis impossible information from the intermediary device 4 via the communication unit 54. When a search instruction for a specific medical image is received from a user via the operation unit 51, the acquisition unit 500 acquires the medical image from the storage unit 53. The output unit 503 outputs the acquired medical image and the lesion analysis impossible information to the client terminal 5b. The display unit 56 of the client terminal 5b displays the specific medical image searched for by the user and the lesion analysis impossible information on the screen (step S50).

[0106] According to the first embodiment, before computer-based lesion detection processing is performed on a medical image, a determination is made as to whether or not to perform computer-based lesion detection processing. In other words, before computer-based lesion detection processing is performed, a filtering process is performed on the medical image to determine whether the medical image is suitable for computer-based lesion detection processing. This allows computer-based lesion detection processing to be performed only on medical images that are suitable for computer-based lesion detection processing. As a result, inaccurate analysis results can be prevented from being provided to users such as doctors, and doctors can reliably avoid overlooking lesions, misdiagnosis, etc.

[0107] Furthermore, according to the first embodiment, before lesion detection processing by a computer is performed, a filtering process is performed on medical images to determine whether the medical images are suitable for lesion detection processing by a computer. In other words, lesion detection processing is not performed on all medical images, but only on specific medical images. This eliminates unnecessary processing in the analysis unit 402, thereby reducing the load on the intermediary device 4 and improving the efficiency of lesion detection processing, etc.

[0108] Furthermore, according to the first embodiment, when a medical image is determined to be unsuitable for computer-based lesion detection processing, the reason and basis for the determination are displayed as lesion analysis unsuitable information I on the client terminal 5b where a doctor or other user refers to the medical image. This allows a doctor or other user to accurately understand whether an abnormality in the lesion was not detected in the computer-based lesion detection processing or whether the medical image is unsuitable for computer-based lesion detection processing. This allows a doctor or other user to make an appropriate diagnosis.

[0109] Second Embodiment In the medical information processing system 100B of the second embodiment, computer-based lesion detection processing is performed by an image analysis device on the cloud. The following description will focus on differences from the first embodiment, and will omit a description of commonalities with the first embodiment. In the description of the second embodiment, components having configurations and functions common to the first embodiment will be denoted by the same reference numerals.

[0110] [Configuration example of medical information processing system 100B] FIG. 9 is a diagram illustrating an example of a schematic configuration of a medical information processing system 100B according to the second embodiment. The medical information processing system 100B includes an in-facility system 10 arranged in a hospital facility and an image analysis device 9 provided on the cloud. The in-facility system 10 includes a hospital information system 1, a radiology information system 2, an imaging device 3, an intermediary device 4, and an image storage device 5. The in-facility system 10 and the image analysis device 9 are communicatively connected to each other via a network N2. Examples of the network N2 include the Internet and a WAN. WAN is an abbreviation for Wide Area Network. The image analysis device 9 may be provided not only on the cloud via the network N2 but also on a network N1 within the medical institution.

[0111] [Example of operation of medical information processing system 100B] The imaging device 3 of the in-facility system 10 receives an examination order from the hospital information system 1 or the like and acquires medical images of the imaging site by imaging the subject. The intermediary device 4 determines whether to perform computer-based lesion detection processing on the medical image acquired by the imaging device 3. For example, the intermediary device 4 determines whether to perform lesion detection processing based on a condition of whether specific image processing has been performed, or determines whether to perform lesion detection processing using a trained model. If the intermediary device 4 determines to perform computer-based lesion detection processing on the medical image acquired by the imaging device 3, it transmits the medical image to the image analysis device 9 on the cloud via the network N2. On the other hand, if the intermediary device 4 determines not to perform computer-based lesion detection processing on the medical image acquired by the imaging device 3, it adds lesion analysis impossible information to the medical image and stores it in the image storage device 5.

[0112] The image analysis device 9 on the cloud functions as the analysis unit 402 of the intermediary device 4 according to the first embodiment. The image analysis device 9 performs a computer-based lesion detection process on medical images acquired from the intermediary device 4 of the in-facility system 10. For example, the image analysis device 9 inputs the acquired medical images into a trained model and acquires medical images with lesion analysis results from the trained model. The image analysis device 9 transmits the acquired medical images with the lesion analysis results to the intermediary device 4 of the in-facility system 10 via the network N2.

[0113] The intermediary device 4 receives medical images with lesion analysis results added thereto from the image analysis device 9 on the cloud, and sends the received medical images to the image storage device 5. The image storage device 5 stores the medical images with lesion analysis results added thereto sent from the intermediary device 4. When a search instruction for a specific medical image is received from a user such as a doctor, the image storage device 5 outputs the medical image to the client terminal 5b. The client terminal 5b displays the specific medical image searched for by the user and the lesion analysis results added to the medical image on the screen of the display unit 56. On the other hand, if lesion detection processing has not been performed on the medical image, the client terminal 5b displays information on the screen indicating the reason and basis why the lesion detection processing for the medical image requested for search was inappropriate.

[0114] The second embodiment can achieve the same effects as the first embodiment. Furthermore, in the past, when computer-based lesion detection processing was performed on a device on the cloud, transmitting all medical images could result in the inclusion of medical images that were not suitable for computer-based lesion detection processing. This could result in unnecessary communication and computational processing associated with the lesion detection processing. In particular, when an image analysis device 9 is installed on the cloud, costs may be incurred based on the amount of data generated by communication and CPU usage fees for the computation of the lesion detection processing. Therefore, uploading all medical images to the cloud could result in unnecessary costs. According to the second embodiment, even when the image analysis device 9 is installed on the cloud, only medical images suitable for computer-based lesion detection processing are transmitted, thereby reducing unnecessary communication and computational processing associated with the lesion detection processing. This avoids unnecessary costs.

[0115] <Third embodiment> In a medical information processing system 100C of the third embodiment, computer-based lesion detection processing is performed by an image analysis device 9 on the cloud, and an image capturing and archiving device 8 is constructed that integrates an imaging device and an image archiving device. Note that the following description will focus on differences from the first embodiment, and descriptions of points in common with the first and second embodiments will be omitted. Note that in the description of the third embodiment, parts that are the same as those in the first embodiment will be described with the same reference numerals.

[0116] [Configuration example of medical information processing system 100C] FIG. 10 is a diagram showing an example of the schematic configuration of a medical information processing system 100C according to the third embodiment. The medical information processing system 100C includes an in-facility system 10 located within a hospital facility and an image analysis device 9 provided on the cloud. The in-facility system 10 includes a medical accounting system 7 and an image capturing and storage device 8. The medical accounting system 7 is a system that integrates the hospital information system 1 and the radiology information system 2 of the first embodiment. The image capturing and storage device 8 is a device that integrates the imaging device 3 and the image storage device 5 of the first embodiment.

[0117] [Configuration example of image capture and archiving device 8] Next, the configuration of the image capturing and archiving device 8 according to the third embodiment will be described. FIG. 11 is a block diagram of the image capturing and archiving device 8 according to the third embodiment. The image capturing and archiving device 8 includes a control unit 80, an operation unit 81, a display unit 82, a storage unit 83, a communication unit 84, and an imaging control unit 85. The control unit 80, the operation unit 81, the display unit 82, the storage unit 83, the communication unit 84, and the imaging control unit 85 are connected to one another via wiring such as a bus 87. The control unit 80 functions as an acquisition unit 800, a determination unit 801, and an output unit 803. The imaging control unit 85 operates as an image acquisition unit and acquires predetermined medical images by capturing images of a subject. A client terminal 8a having a display unit 86 is connected to the image capturing and archiving device 8.

[0118] [Configuration example of image analysis device 9] Next, the configuration of the image capturing and storing device 8 according to the third embodiment will be described. Fig. 12 is a block diagram of the image capturing and storing device 8 according to the third embodiment. The image analyzing device 9 includes a control unit 90, an operation unit 91, a storage unit 93, and a communication unit 94. The control unit 90, the operation unit 91, the storage unit 93, and the communication unit 94 are connected to each other via wiring such as a bus 97. The control unit 90 functions as an acquisition unit 900, an analysis unit 902, and an output unit 903.

[0119] [Example of operation of medical information processing system 100C] The imaging control unit 85 of the image capturing and archiving device 8 receives an examination order from the medical accounting system 7 or the like and captures medical images of the captured region by capturing images of the subject. The medical images captured by the imaging control unit are stored in the archiving unit 83. When a user, such as a doctor, requests lesion detection processing of a specific image, the specific image is selected on the screen displayed on the display unit 82 and an operation to request lesion detection processing is performed. The determination unit 801 reads the medical image selected by the user from the archiving unit 83 and determines whether to perform computer-based lesion detection processing on the read medical image. For example, the determination unit 801 determines whether to perform lesion detection processing based on a condition of whether specific image processing has been performed, or whether to perform lesion detection processing using a trained model. If the output unit 803 determines to perform computer-based lesion detection processing on the medical image, it transmits the medical image to the image analysis device 9 on the cloud via the network N2. On the other hand, if the determination unit 801 determines not to perform computer-based lesion detection processing on the medical image, it adds lesion analysis impossible information to the medical image and stores it in the archiving unit 83.

[0120] The image analysis device 9 functions as the analysis unit 402 of the intermediary device 4 of the first embodiment. The image analysis device 9 performs a computer-based lesion detection process on medical images acquired from the image capture and storage device 8 of the in-facility system 10. For example, the image analysis device 9 inputs the acquired medical images into a trained model and acquires medical images with lesion analysis results added from the trained model. The image analysis device 9 transmits the acquired medical images with lesion analysis results added to the image capture and storage device 8 of the in-facility system 10 via the network N2.

[0121] The communication unit 84 of the image capture and storage device 8 receives medical images with lesion analysis results added from the image analysis device 9 on the cloud. The acquisition unit 800 stores the received medical images with lesion analysis results added in the storage unit 83. The output unit 803 outputs the medical images with lesion analysis results added to the client terminal 5b. The display unit 56 of the client terminal 5b displays on the screen of the display unit 56 the specified medical image searched by the user and the lesion analysis results added to the medical image. On the other hand, if lesion detection processing has not been performed on the medical image, the client terminal 5b displays on the screen information indicating the reason and basis why the lesion detection processing for the medical image requested for search was inappropriate.

[0122] According to the third embodiment, the same effects as those of the first embodiment can be achieved. Furthermore, in the past, when computer-based lesion detection processing was performed on a device on the cloud, transmitting all medical images could result in the inclusion of medical images that were not suitable for computer-based lesion detection processing. This could result in unnecessary communication, load, and the like. In particular, when an image analysis device 9 is installed on the cloud, usage fees may be incurred depending on the amount of processing, time, and the like. Therefore, uploading all medical images to the cloud could result in unnecessary costs. According to the third embodiment, even when the image analysis device 9 is installed on the cloud, only medical images that are suitable for computer-based lesion detection processing are transmitted, thereby reducing unnecessary communication and processing. This can avoid unnecessary costs.

[0123] While the preferred embodiments of the present disclosure have been described in detail above with reference to the accompanying drawings, the technical scope of the present disclosure is not limited to such examples. Furthermore, various modifications and improvements will naturally fall within the technical scope of the present disclosure, provided that they are within the scope of the technical ideas described in the claims of those skilled in the art. For example, medical images suitable for computer-based lesion detection processing may be limited to medical images of adults. Whether a medical image is not of a child can be determined by calculating the age information at the time of imaging from the examination date, series date, image date, or acquisition date in the tag information attached to the medical image according to the DICOM standard, and the patient's date of birth, and then judging from the age information at the time of imaging. [Explanation of symbols]

[0124] 1. Hospital Information System 2 Radiology Information System 3. Imaging equipment 4. Intermediary device (medical information processing device) 5 Image Archive 40 Control Unit 56 Display section 100A, 100B, 100C Medical Information Processing System 400 Acquisition Department 401 Judgment section 402 Analysis Department 403 Output section I. Information on lesions that cannot be analyzed

Claims

1. an image acquisition unit for acquiring medical images; a determination unit that determines whether or not to perform a lesion detection process by computer on the medical image, A medical information processing device wherein the determination unit determines whether to perform the lesion detection processing based on at least a condition of whether specific image processing has been performed, or determines whether to perform the lesion detection processing using a trained model.

2. an output unit that outputs the medical image based on the determination of whether or not to make the determination; The medical information processing device according to claim 1.

3. an acquisition unit that acquires information on whether the specific image processing has been performed by using supplementary information of the medical image or image recognition of the medical image; the determination unit determines whether to perform the lesion detection process based on at least a condition of whether or not the specific image processing has been performed, using the acquired information on whether or not the specific image processing has been performed. The medical information processing device according to claim 1.

4. an acquisition unit that acquires information on whether the specific image processing has been performed from information included in a hospital information system or information included in a radiology department information system; the determination unit determines whether to perform the lesion detection process based on at least a condition of whether or not the specific image processing has been performed, using the acquired information on whether or not the specific image processing has been performed. The medical information processing device according to claim 1.

5. The trained model is a trained model using a dataset in which at least a medical image that has undergone specific image processing or a medical image that has not undergone image processing is paired with a correct answer label.

3. The medical information processing device according to claim 1.

6. The specific image processing is at least one processing selected from a group of processing including up / down / left / right inversion processing, rotation processing, black / white inversion processing, For Processing processing, bone attenuation processing, temporal subtraction processing, energy subtraction processing, outside-irradiation-field darkening processing, frequency processing above a certain intensity, addition of digital markers, and addition of annotations.

3. The medical information processing device according to claim 1.

7. The determination unit determines whether or not to perform computer-based lesion detection processing on the medical image based on at least one of the following conditions: whether or not the medical image is taken under standard imaging conditions; whether or not the medical image satisfies specific image quality conditions; whether or not a detection target site to be detected in the lesion detection processing is missing; whether or not the medical image contains a foreign object that is not a detection target; and whether or not the medical image is an image taken by a modality other than a specific modality.

3. The medical information processing device according to claim 1.

8. The determination unit determines whether to perform the lesion detection process using a trained model that uses, as training data, medical images including at least one of the following information: whether the imaging conditions are standard; whether the medical image satisfies specific image quality conditions; whether a detection target site to be detected in the lesion detection process is missing; whether the medical image includes a foreign object that is not a detection target; and whether the medical image is an image captured by a modality other than a specific modality.

3. The medical information processing device according to claim 1.

9. Whether or not the medical image is under the standard imaging conditions is determined based on whether or not the tilt of the medical image satisfies the conditions, whether or not the medical image is a portable imaging image, whether or not the medical image is a magnified imaging image, whether or not the medical image is a spot imaging image, whether or not the imaging position of the subject of the medical image satisfies the conditions, whether or not the dose information of the medical image satisfies the conditions, and whether or not the medical image is an image in which moire occurs. The medical information processing device according to claim 7 or 8.

10. Whether the medical image satisfies a specific image quality condition is at least one of whether the medical image is a film digitized image, whether the resolution of the medical image satisfies the condition, and whether lossy compression processing has been performed. The medical information processing device according to claim 7 or 8.

11. Whether or not the detection target portion to be detected in the lesion detection process is missing is at least one of whether or not the medical image is a lung field missing image, and whether or not the imaging target portion is an image cut off outside the medical image. The medical information processing device according to claim 7 or 8.

12. Whether or not a foreign object that is not a detection target is included in the medical image is determined based on at least one of whether or not the medical image contains a foreign object other than a living body, and whether or not a large lesion is present in the medical image. The medical information processing device according to claim 7 or 8.

13. Whether or not the medical image is an image captured by a modality other than the specific modality is determined by whether or not the medical image is at least one of a tomographic image, an ultrasound image, and a video image. The medical information processing device according to claim 1.

14. The determination unit determines whether or not to perform a computer-based lesion detection process on the medical image based on whether or not the medical image has been captured under specific imaging conditions other than the imaging direction or the imaging region. The medical information processing device according to claim 1.

15. an acquisition unit that acquires at least one piece of information from tag information of the medical image, including whether or not the specific image processing has been performed, whether or not the medical image is taken under the standard imaging conditions, whether or not the medical image satisfies specific image quality conditions, and whether or not the medical image is an image taken by a modality other than a specific modality; the determination unit determines whether or not to perform the lesion detection process using information acquired from the tag information. The medical information processing device according to claim 7.

16. an acquisition unit that acquires at least one piece of information by image recognition of the medical image, including whether the specific image processing has been performed, whether the imaging conditions are standard, whether the medical image satisfies specific image quality conditions, whether a detection target site to be detected by the lesion detection processing is missing, whether a foreign object that is not a detection target is included in the medical image, and whether the medical image is an image captured by a modality other than a specific modality; the determination unit determines whether or not to perform the lesion detection process using information recognized by the image recognition. The medical information processing device according to claim 7.

17. an acquisition unit that acquires at least one piece of information, by image recognition of the medical image, whether or not a detection target site that is a detection target of the lesion detection processing is missing, and whether or not a foreign object that is not a detection target is included in the medical image; The medical information processing device according to claim 7.

18. the determination unit makes a determination corresponding to the type of the lesion detection processing, The medical information processing device according to claim 1.

19. the output unit outputs information related to the determination when the determination unit determines that the lesion detection process should not be performed. The medical information processing device according to claim 2.

20. the output unit outputs a reason for the determination when the determination unit determines that the lesion detection process should not be performed. The medical information processing device according to claim 2.

21. the output unit, when the determination unit determines that the lesion detection process should not be performed, outputs a medical image indicating the basis for the determination. The medical information processing device according to claim 2.

22. an image acquisition unit for acquiring medical images; a determination unit that determines whether or not to perform a lesion detection process by computer on the medical image, A medical information processing system in which the judgment unit determines whether to perform the lesion detection processing based on at least the condition of whether specific image processing has been performed, or determines whether to perform the lesion detection processing using a trained model.

23. an image acquisition step of acquiring a medical image; a determination step of determining whether or not to perform lesion detection processing by a computer on the medical image, A medical information processing method in which the judgment step determines whether to perform the lesion detection processing based on at least the condition of whether specific image processing has been performed, or determines whether to perform the lesion detection processing using a trained model.

24. Computer, an image acquisition unit for acquiring medical images; a determining unit that determines whether or not to perform computer-based lesion detection processing on a medical image; The program in which the determination unit determines whether to perform the lesion detection processing based on at least a condition of whether specific image processing has been performed, or determines whether to perform the lesion detection processing using a trained model.

Images