WO2021066039A1 - Medical information processing program, and medical information processing device - Google Patents

Abstract

This medical information processing device includes a control unit that acquires target data including data of a medical image relating to a person to be diagnosed, and information about the medical image imaging timing (S1). The control unit extracts, as similar case data, a case data item that includes a medical image similar to the medical image included in the target data, from a plurality of case data items (S3). The control unit matches the time axis of the medical image imaging timing between a plurality of data items (S7, S10, S14). The control unit outputs medical information based on the similar case data, with the time axes being matched (S7, S11, S15, S18, S22).

Description

Medical information processing program and medical information processing equipment

The present disclosure relates to a medical information processing program and a medical information processing device for outputting medical information useful for diagnosis based on case data similar to the data of a person to be diagnosed among a plurality of case data.

Conventionally, various techniques for presenting medical information useful for diagnosis to users have been proposed. For example, the ophthalmic information processing apparatus disclosed in Patent Document 1 facilitates follow-up observation of an eye to be inspected by matching layer thickness information based on a plurality of OCT data obtained on different days.

JP-A-2017-104535

It is also conceivable to extract case data similar to the data of the person to be diagnosed from a plurality of case data including a plurality of medical images taken at different timings, and output medical information based on the extracted case data. In this case, since the transition of the disease state of the patient is grasped from the plurality of medical images included in the extracted case data, the transition of the disease of the diagnosis target person is appropriately predicted by the output medical information. ..

However, the multiple medical images included in the case data and the medical images of the person to be diagnosed are both taken at various timings. Therefore, if the medical information is not output in consideration of the timing at which each medical image is taken, the diagnosis by the doctor may not be appropriately assisted.

The present disclosure is to provide a medical information processing program and a medical information processing device capable of presenting more useful medical information to a user.

The medical information processing program provided by the typical embodiment in the present disclosure is medical information processing that outputs medical information useful for diagnosis based on case data similar to the data of a person to be diagnosed among a plurality of case data. It is a medical information processing program executed in the device, and each of the plurality of case data includes data of a plurality of medical images taken at different timings and information on the shooting timing of each medical image. When the medical information processing program is executed by the control unit of the medical information processing apparatus, the data of at least one medical image relating to the diagnosis target person and the imaging timing of the medical image are determined. From the target data acquisition step for acquiring target data including information and the plurality of case data, the case data including a medical image similar to at least one medical image included in the target data is extracted as similar case data. A similar case data extraction step, and a time axis matching step of matching the time axis of the acquisition timing of the medical image between the similar case data, the target data, and at least one of the plurality of similar case data. , The medical information output step of outputting medical information based on the similar case data in a state where the time axes are matched, is executed by the medical information processing apparatus.

The medical information processing apparatus provided by the typical embodiment in the present disclosure is medical information processing that outputs medical information useful for diagnosis based on case data similar to the data of a person to be diagnosed among a plurality of case data. It is an apparatus, and each of the plurality of case data is stored in a database in a state including data of a plurality of medical images taken at different timings and information on the timing of taking each medical image. The control unit of the medical information processing apparatus includes a target data acquisition step of acquiring at least one medical image data relating to the diagnosis target person and target data including information on the imaging timing of the medical image, and the plurality of case data. A similar case data extraction step for extracting the case data including a medical image similar to at least one medical image included in the target data as similar case data, the similar case data and the target data, and a plurality of the similar case data. Medical information based on the similar case data in a state where the time axis matching step of matching the time axis of the acquisition timing of the medical image and the time axis are matched between at least one of the similar case data of the above. To output the medical information output step, and to execute.

According to the medical information processing program and medical information processing method in the present disclosure, more useful medical information is presented to the user.

The medical information processing program illustrated in this disclosure is executed in a medical information processing device. The medical information processing device outputs medical information useful for diagnosis based on case data similar to the data of the person to be diagnosed among a plurality of case data. Each of the plurality of case data is stored in the database in a state including the data of the plurality of medical images taken at different timings and the information of the shooting timing of each medical image. The control unit of the medical information processing apparatus executes a target data acquisition step, a similar case data extraction step, a time axis matching step, and a medical information output step. In the target data acquisition step, the control unit acquires target data including at least one medical image data regarding the diagnosis target person and information on the timing of taking the medical image. In the similar case data extraction step, the control unit extracts case data (similar case data) including a medical image similar to at least one medical image included in the target data from the plurality of case data. In the time axis matching step, the control unit matches the time axis of the acquisition timing of the medical image between the similar case data, the target data, and at least one of the plurality of similar case data. In the medical information output step, the control unit outputs medical information based on the extracted case data in a state where the time axes are matched.

According to the medical information processing apparatus exemplified in the present disclosure, the time axis of the acquisition timing of the medical image is set between a plurality of data (between similar case data and target data, and at least one of a plurality of similar case data). Medical information is output in the matched state. Therefore, the user (for example, a doctor or the like) who confirms the output medical information can appropriately predict the transition of the disease of the diagnosis target person after grasping the timing when each medical image is taken. Therefore, more useful medical information is presented to the user.

The method for extracting similar case data can be set as appropriate. For example, the control unit extracts a value indicating the feature amount of each medical image (for example, SIFT (Scale Invariant Feature Transfer Transfer)), and selects a medical image having a small difference in the feature amount between the target data and the medical image. The including case data may be extracted as similar case data. In addition, the control unit may acquire the similarity of the medical image by inputting the medical image into the mathematical model trained by the machine learning algorithm, and extract similar case data based on the acquired result. Further, when the image quality of the medical image included in the case data is less than the threshold value, the control unit may exclude it from the extraction target or warn the user that the image quality is low.

In the time axis matching step, the control unit may match the time axis of the medical image imaging timing between the plurality of data according to the instruction input from the user. In this case, the user can output appropriate medical information to the medical information processing apparatus in a state where the time axis between the plurality of data matches a desired state.

The specific method for matching the time axis according to the instruction from the user can be appropriately selected. For example, a time axis indicating the acquisition timing of each medical image of a plurality of data may be displayed on the display unit. The user may input an instruction for adjusting the time axis to the medical information processing apparatus while confirming the time axis indicating the imaging timing. The control unit may match the time axis according to the input adjustment instruction. In this case, the control unit may display the medical image on the display unit together with the time axis. The user may also enter an instruction to specify a criterion for matching the time axis. For example, the control unit may input an instruction to specify a medical image whose imaging timing matches on the time axis as a reference image among the medical images of the plurality of data. In this case, the control unit may match the time axes so that the imaging timings of the medical images designated as the reference images match.

In the time axis matching step, the control unit may match the time axis by matching the imaging timings of the medical images having the highest degree of similarity among the medical images included in each of the plurality of data. The disease states in the two most similar medical images are likely to be similar. Therefore, by matching the time axes so that the acquisition timings of the two medical images having the highest degree of similarity match, the user can more appropriately grasp the transition of the disease from the medical information.

The case data may include information (treatment information) regarding the treatment (eg, medication, surgery, treatment, etc.) performed on the patient. In this case, the medical information processing apparatus can output more appropriate medical information based on the treatment information included in the case data.

The specific content of treatment information can be selected as appropriate. For example, treatment data includes information indicating whether or not treatment has been performed on the patient, information indicating the content of treatment performed on the patient, and treatment in which treatment has been performed or started on the patient. It may contain at least one of the timing information.

In the target data acquisition step, the control unit is one of a plurality of case data groups classified in the plurality of case data according to at least one of the presence or absence of treatment indicated by the treatment information and the content of the performed treatment. From a plurality of case data groups, case data including a medical image similar to at least one medical image included in the target data (hereinafter, referred to as “similar case data”) may be extracted. In this case, medical information is output after appropriately extracting case data according to at least one of the presence or absence of treatment and the content of treatment. Therefore, the user can appropriately predict the transition of the disease of the diagnosis target person according to the content of the treatment and the like.

Note that the control unit may extract similar case data from one case data group among a plurality of case data groups. Further, the control unit may extract at least one similar case data from each of the plurality of case data. For example, by extracting similar case data from each of the case data group of treated patients and the case data group of untreated patients, the transition of the disease according to the presence or absence of treatment can be appropriately performed. is expected. In addition, by extracting similar case data from each of a plurality of treatment data groups having different treatment contents, the transition of the disease according to the treatment contents can be appropriately predicted.

The treatment information included in the case data may include information indicating the treatment timing when the treatment is executed or started. In the time axis matching step, the control unit may match the time axis by matching the treatment timing between the plurality of data. In this case, the transition of the diagnosed subject due to the treatment of the disease is appropriately predicted based on the treatment timing.

The timing of treatment for the diagnosis target person may be the timing of the treatment actually performed for the diagnosis target person, or the timing of the treatment assuming that the diagnosis target person will be treated in the future.

It is also possible to change the specific method for matching the time axes of multiple data. For example, the control unit may specify the timing (change timing) at which the change in the feature amount of the plurality of medical images becomes equal to or greater than the threshold value for each of the plurality of data. The control unit may match the time axis by matching the change timings of the plurality of data. In this case, the transition of the disease is appropriately predicted based on the change timing. For example, the control unit identifies the change timing at which the change in the progress of the disease (progress of deterioration) exceeds the threshold value based on the feature amounts of a plurality of medical images, and matches the change timing of the plurality of data. You may let me. In this case, the user can easily compare a plurality of data based on the timing when the progress of the disease suddenly deteriorates.

When extracting similar case data from a plurality of case data, the control unit determines the degree of similarity between the medical image included in the case data before or after the treatment / change timing and the medical image of the target data. Similar case data may be extracted by comparing. In this case, similar case data is extracted more appropriately after considering the treatment / change timing.

In the similar case data extraction step, when the target data includes a plurality of medical images, the control unit captures a plurality of images at an imaging interval in which the difference from the imaging interval of the plurality of medical images in the target data is equal to or less than the threshold. From the case data including the medical images of the above, case data in which a plurality of medical images are similar to each other may be extracted as similar case data. In this case, the plurality of medical images of the extracted similar case data include medical images taken at intervals close to the shooting intervals of the plurality of medical images of the target data. Therefore, the user can more appropriately predict the transition of the disease based on the case data in which the imaging intervals of the plurality of medical images are close and the time axes are the same. The threshold value of the difference between the imaging intervals of the medical image of the target data and the medical image of the case data may be set in advance or may be set according to an instruction input by the user.

When the target data includes a plurality of medical images, the control unit can extract similar case data by various methods. For example, the control unit sets one or a plurality of medical images to be determined for similarity with the case data among a plurality of medical images included in the target data according to an instruction input from the user. You may. The control unit may extract similar case data by comparing the similarity between the medical image to be determined and the medical image of the case data.

In the medical information output step, the control unit displays the medical image included in each of the plurality of data and the time axis indicating the shooting timing of the medical image as medical information on the display unit in a state where the time axes are matched. It may be displayed. In this case, the user can appropriately grasp the imaging timing of the plurality of medical images displayed on the display unit in a state where the time axes of the plurality of data are matched. Therefore, the user can more easily predict the transition of the disease.

The control unit may display the medical image included in each of the plurality of similar case data on the display unit together with the time axis. Further, the control unit may display the medical image included in the similar case data and the medical image included in the target data on the display unit together with the time axis. In this case, the user can appropriately compare the medical image of the target data and the medical image of the similar case data after grasping the imaging timing on the time axis.

The control unit is based on the medical image included in the similar case data, the medical image taken after the same timing as the shooting timing of the medical image of the target data on the matched time axis, and the medical image of the target data. Therefore, a predicted image of the person to be diagnosed may be generated. In the medical information output step, the control unit may display the predicted image generated in a state where the time axes are matched on the display unit as medical information. In this case, an appropriate predicted image is generated and displayed based on the medical images of the similar case data and the target data, taking into consideration the imaging timing. Therefore, the user can more appropriately predict the transition of the disease.

The control unit may further execute the progress acquisition step and the graph generation step. In the progress acquisition step, the control unit acquires information on the progress of the disease shown in each of the plurality of medical images. In the graph generation step, the control unit displays a progress transition graph showing the transition of the progress in each of the similar case data and the target data, or a plurality of similar case data, in a state where the time axis between the plurality of data is matched. Generate. In the medical information output step, the control unit may display the progress transition graph generated in a state where the time axes are matched on the display unit as medical information. In this case, the user can easily grasp the transition of the disease progression in the plurality of data. Further, since the progress transition graph is generated in a state where the time axes of the plurality of data match, the user can appropriately compare the progress transition graphs of the plurality of data.

At least one of the plurality of case data may include heterogeneous data acquired by a device of a type different from the medical imaging device that captured the medical image included in the target data regarding the diagnosis target person. In the medical information output step, the control unit may output the heterogeneous data as medical information when the extracted similar case data includes the heterogeneous data. The heterogeneous data is, for example, an image taken by a different type of imaging device from the medical imaging apparatus that captured the medical image included in the target data (that is, the medical image referred to when extracting similar case data). It may be. In addition, the heterogeneous data is the test result acquired by a test device that tests the patient (for example, in the field of ophthalmology, at least one of the test results such as visual acuity, axial length, intraocular pressure, and visual field of the eye to be examined). It may be. Further, the heterogeneous data may be displayed together with the medical image included in the similar case data, or the heterogeneous data may be output independently (for example, display).

In this case, the user can make a diagnosis after confirming the heterogeneous data included in the similar case data, so that the transition of the disease of the person to be diagnosed can be predicted more appropriately. In addition, even if the heterogeneous data regarding the diagnosis target person has not been acquired yet, the user can perform the diagnosis of the diagnosis target person more appropriately by checking the heterogeneous data included in the similar case data. it can.

It is a block diagram which shows the schematic structure of the medical information processing system 1. It is a flowchart of medical information processing executed by the medical information processing apparatus 20. It is explanatory drawing for demonstrating an example of the method of extracting case data 50 similar to target data 40 from a plurality of case data 50. It is a figure which shows an example of the display screen which displayed the medical image of a plurality of data together with the time axis T. It is explanatory drawing for demonstrating an example of the generation method of the prediction image 60. It is a figure which shows an example of the progress transition graph.

Hereinafter, one of the typical embodiments in the present disclosure will be described with reference to the drawings. In the present embodiment, medical information processing including a server 10, a plurality of medical information processing devices 20 used at each base, and a medical image capturing device 30 for supplying data including medical images to the medical information processing device 20. The system 1 is illustrated. However, the configurations of the medical information processing system and the medical image processing apparatus are not limited to the configurations exemplified in this embodiment. For example, in the present embodiment, the storage device of the server 10 is used as a database for storing a plurality of case data. However, when the case data is stored in the storage device 23 of the medical information processing device 20, the server 10 can be omitted. Further, the server 10 or the medical imaging device 30 may function as a medical information processing device. Further, a plurality of devices (for example, two or more such as a server, a terminal device (personal computer or mobile terminal, etc.), and a medical imaging device) may cooperate to function as a medical information processing device.

The system configuration of the medical information processing system 1 in the present embodiment will be described with reference to FIG. As described above, the medical information processing system 1 of the present embodiment includes a server 10 and a plurality of medical information processing devices 20 used at each base (for example, a hospital, a medical examination facility, etc.). FIG. 1 illustrates a medical information processing device 20A used at the base A and a medical information processing device 20B used at the base B.

The server 10 provides various data and the like to the connected device (medical information processing device 20 in this embodiment). In this embodiment, a server (so-called cloud server) of a manufacturer that provides a cloud service is used as the server 10. However, it goes without saying that a server other than the cloud server may be used. The server 10 includes a control unit 11 that performs various processing controls and a communication I / F 14. The control unit 11 includes a CPU 12 which is a controller that controls control, and a storage device 13 that can store programs, data, and the like. In the present embodiment, the storage device 13 of the server 10 is used as a database for storing case data described later. The communication I / F 14 connects the server 10 to an external device (for example, the medical information processing device 20) via the network 5 (for example, the Internet).

The medical information processing device 20 is used by users at each base (for example, doctors and laboratory technicians). Although the medical information processing device 20 of the present embodiment is a personal computer, a mobile terminal such as a smartphone or a tablet terminal may be used as the medical information processing device. The medical information processing device 20 includes a control unit 21 that performs various control processes and a communication I / F 24. The control unit 21 includes a CPU 22 which is a controller that controls control, and a storage device 23 that can store programs, data, and the like. The storage device 23 stores a medical information processing program for executing various processes described later. Further, the communication I / F 24 connects the medical information processing device 20 to an external device (for example, a server 10) via the network 5. For example, the medical information processing apparatus 20 receives (acquires) case data from the server 10. Further, the medical information processing apparatus 20 transmits (outputs) case data to the server 10.

The medical information processing device 20 is connected to the operation unit 25 and the display unit 26. The operation unit 25 is operated by the user in order for the user to input various instructions to the medical information processing apparatus 20. For the operation unit 25, for example, at least one of a keyboard, a mouse, a touch panel, and the like can be used. A microphone or the like for inputting various instructions may be used together with the operation unit 25 or instead of the operation unit 25. The display unit 26 is a device (for example, a monitor or a projector) capable of displaying various images.

At least one of the plurality of medical information processing devices 20 can exchange data (for example, medical image data, etc.) with one or a plurality of medical image capturing devices 30 that capture a medical image of a patient. .. The method by which the medical information processing apparatus 20 exchanges data and the like with the medical imaging apparatus 30 can be appropriately selected. For example, the medical information processing apparatus 20 exchanges data and control signals with the medical imaging apparatus 30 by at least one of wired communication, wireless communication, a detachable storage medium (for example, a USB memory), and the like. May be good.

Various devices can be used for the medical imaging device 30. As an example, the medical imaging device 30 used in the present embodiment includes an OCT device capable of acquiring a tomographic image and a frontal image of the tissue of the eye to be inspected (the fundus in the present embodiment). However, an ophthalmologic imaging device other than the OCT device (for example, at least one of a fundus camera, a scanning laser ophthalmoscope (SLO), a corneal shape measuring device, and the like) may be used. In addition, a medical imaging device that photographs the tissue of a patient other than the eye to be inspected may be used.

The medical imaging device 30 includes a control unit 31 that performs various control processes and an imaging unit 35. The control unit 31 includes a CPU 32, which is a controller that controls control, and a storage device 33 that can store programs, data, and the like. The imaging unit 35 includes various configurations necessary for the medical imaging apparatus 30 to capture a medical image of a patient. For example, when an OCT device is used as the medical imaging device 30, the imaging unit 35 includes an OCT light source, a scanning unit for scanning OCT light, an optical system for irradiating the eye to be inspected with OCT light, and an eye to be inspected. A light receiving element or the like that receives light reflected by the tissue is included.

An example of medical information processing in the present embodiment will be described with reference to FIGS. 2 to 6. The medical information processing illustrated in FIG. 2 is executed by the CPU 22 of the medical information processing device 20 according to the medical information processing program stored in the storage device 23 of the medical information processing device 20. However, as described above, the medical information processing may be executed by the control unit of another device (for example, the CPU 12 of the server 10 or the CPU 32 of the medical imaging device 30). Further, the control units of a plurality of devices (for example, the CPU 22 of the medical information processing device 20 and the CPU 12 of the server 10) may cooperate to execute medical information processing.

First, the CPU 22 acquires the target data 40 regarding the diagnosis target person (S1). As shown in FIGS. 3 and 4, in the target data 40 of the present embodiment, at least one medical image 41 taken by the medical imaging apparatus 30 and each medical image 41 were taken for the diagnosis target person. Information on the shooting timing (in the present embodiment, information on the shooting date and time) is included. The CPU 22 may acquire the target data 40 from the medical imaging apparatus 30 via communication, a detachable storage medium, or the like. Further, the CPU 22 uses the target data of the past target data of the diagnosis target person already stored in the storage device 23, the medical image 41 newly captured by the medical image capturing device 30, and the imaging timing information. It may be acquired as 40. Although the details will be described later, the medical information processing apparatus 20 extracts case data similar to the target data 40 as similar case data from a plurality of case data stored in the database.

Here, a plurality of case data 50 stored in the database will be described with reference to FIGS. 3 and 4. In the present embodiment, the storage device 13 of the server 10 is used as a database for storing a plurality of case data 50. However, the database that stores the case data 50 may be another storage device (for example, the storage device 23 of the medical information processing device 20). The case data 50 includes a plurality of medical images 51 captured at different timings by the medical imaging apparatus 30, and information on the imaging timing of each medical image 51. The plurality of case data 50 are stored in the database.

Further, as shown in FIG. 4, the case data 50 of the present embodiment includes information regarding the treatment performed on the patient (hereinafter, referred to as “treatment information”). In the present embodiment, the treatment information includes the presence or absence of treatment for the patient, the timing when the treatment is executed or started (hereinafter referred to as "treatment timing"), and the content of the treatment performed (for example, the type of surgery performed). , At least one of the type of drug administered, the method of administration, the type of treatment performed, etc.).

Each of the plurality of case data 50 stored in the database is classified into one of the plurality of case data groups according to the presence or absence of treatment indicated by the treatment information and the content of the performed treatment. That is, in the present embodiment, each of the plurality of case data 50 is classified into either a case data group of a patient who has not been treated and a case data group of a patient who has been treated. Furthermore, the case data group of the treated patients is further subdivided according to the content of the treatment performed. Although the details will be described later, the medical information processing apparatus 20 can also extract case data 50 similar to the target data 40 according to the classification of the case data group.

In addition, at least one of the plurality of case data stored in the database includes data related to the patient (hereinafter,) acquired by a device of a type different from the medical image capturing device 30 that captures the medical image 41 of the target data 40. , "Different data") may be included. The heterogeneous data includes at least an imaging device (for example, a fundus camera, a scanning laser ophthalmoscope (SLO), a corneal shape measuring device, etc.) different from the medical imaging apparatus 30 (OCT apparatus in the present embodiment) described above. The data of the image taken by either) may be included. In addition, the heterogeneous data includes test result data (for example, at least one test result data such as visual acuity, axial length, intraocular pressure, visual field, etc.) acquired by a test device that tests a patient. It may be.

Return to the explanation in Fig. 2. The CPU 22 uses one or a plurality of medical images 41 as a reference for extracting the case data 50 from the one or a plurality of medical images 41 included in the target data 40 acquired in S1 as a reference image. Set (S2). Although details will be described later, the medical information processing apparatus 20 of the present embodiment has a medical image 51 similar to at least one medical image 41 included in the target data 40 from among a plurality of case data 50 stored in the database. Case data 50 including the above is extracted as similar case data. In S2, from the medical image 41 in the target data 40, the medical image 41 for determining the degree of similarity with the medical image 51 in the case data 50 is set as a reference image.

The specific method for setting the reference image in S2 can be appropriately selected. First, when the target data 40 acquired in S1 includes only one medical image 41, the CPU 22 sets one medical image 41 in the target data 40 as the reference image. When the target data 40 includes a plurality of medical images 41, the CPU 22 uses one or a plurality of medical images 41 selected by the instruction as the reference image based on the reference image selection instruction input from the user. It may be set. When the target data 40 includes a plurality of medical images 41, the CPU 22 may automatically set one or a plurality of medical images 41 as reference images in the order of newest imaging timing.

Next, the CPU 22 extracts one or more case data 50 similar to the target data 40 from the plurality of case data 50 stored in the database as similar case data (S3). Specifically, the CPU 22 extracts the case data 50 including the medical image 51 similar to the reference image set in the target data 40 as the similar case data.

As an example, in S3 of the present embodiment, the CPU 22 acquires a value indicating the feature amount of each of the reference image and the plurality of medical images 51 in the case data 50. In this embodiment, SIFT (Scale Invariant Feature Transfer Transfer) is adopted as a value indicating the feature amount of the image. However, feature quantities other than SIFT may be acquired. The CPU 22 extracts the case data 50 including the medical image 51 having a small difference in the feature amount from the reference image as similar case data. For example, the CPU 50 may extract case data 50 including a medical image 51 in which the difference in feature amounts is equal to or less than a threshold value. Further, the CPU 50 may extract one or more case data 50 in ascending order of the difference between the feature amounts of the reference image and the medical image 51.

With reference to FIG. 3, an example of a method for extracting similar case data when a plurality of reference images 41A and 41B are set in the target data 40 will be described. In the example shown in FIG. 3, two medical images out of the plurality of medical images 41 included in the target data 40 are set as the reference images 41A and 41B. In this case, the CPU 22 specifies the shooting intervals D1 of the plurality of reference images 41A and 41B based on the shooting timing information included in the target data 40. Next, the CPU 22 identifies the case data 50 including the plurality of medical images 51 taken at intervals of the plurality of reference images 41A and 41B with an interval D1 which is equal to or less than the threshold value among the plurality of case data 50. From the identified case data 50, the CPU 22 has a difference in imaging interval between the reference images 41A and 41B being less than or equal to the threshold value and a high degree of similarity with the reference images 41A and 41B (that is, a small difference in feature amount). ) Case data 50 including a plurality of medical images 41 is extracted as similar case data. The threshold value for comparing the shooting intervals may be set in advance or may be set according to an instruction from the user.

In the example shown in FIG. 3, the difference between the imaging interval D2 of the two medical images 51A and 51B included in the case data 50A and the imaging interval D1 of the two reference images 41A and 41B is equal to or less than the threshold value. Further, the similarity between the two medical images 51A and 51B and the two reference images 41A and 41B is high (that is, the similarity between the medical image 51A and the reference image 41A and the similarity between the medical image 51B and the reference image 41B are high. , Both are expensive). Therefore, the CPU 22 extracts the case data 50A as similar case data. On the other hand, although the two medical images 51X and 51Y included in the case data 50B have a high degree of similarity with the two reference images 41A and 41B, the two medical images 51X and 51Y have an imaging interval D3 and two. The difference between the shooting intervals D1 of the reference images 41A and 41B is larger than the threshold value. Therefore, the CPU 22 does not extract the case data 50B as similar case data. By performing the above processing, the extracted similar case data includes medical images 51A and 51B taken at intervals close to the imaging intervals of the plurality of reference images 41A and 41B of the target data 40. ..

When there is only one reference image set in the target data 40, the CPU 22 extracts the case data 50 including the medical image 51 having a high degree of similarity to the one reference image as the similar case data.

Further, in S3 of the present embodiment, the CPU 22 can extract similar case data from one or a plurality of case data groups among a plurality of case data groups classified by treatment information. For example, when an instruction to extract similar case data from a case data group of a patient who has been treated with a specific content is input, the CPU 22 is one from the case data group corresponding to the content of the instructed treatment. Or extract multiple similar case data. The CPU 22 can also extract one or more similar case data from each of the plurality of case data groups. For example, the CPU 22 can also extract each of the similar case data of the treated patient and the similar case data of the untreated patient. The CPU 22 can also extract a plurality of similar case data having different contents of the performed treatment.

Further, in S3 of the present embodiment, the CPU 22 uses the medical image 51 before or after the treatment timing indicated by the treatment information among the plurality of medical images 51 included in each case data 50, and the reference image of the target data 40. It is also possible to extract similar case data by comparing the similarities of. For example, when the reference image of the target data 40 is the medical image 41 before the treatment, the CPU 22 compares the similarity between the medical image 51 before the treatment timing and the reference image and extracts similar case data. As a result, case data 50 in which the state of the disease before treatment is close to the state of the disease of the diagnosed subject is appropriately extracted.

It is also possible to change the extraction method of similar case data. For example, the CPU 22 acquires the similarity of the medical images 41 and 51 by inputting the medical images 41 and 51 into the mathematical model trained by the machine learning algorithm, and extracts similar case data based on the acquired results. You may. Further, when the image quality of the medical image 51 included in the case data 50 is less than the threshold value, the CPU 22 may exclude the case data 50 from the extraction target, or may exclude the extracted similar case data image. The user may be warned that the quality is low.

Return to the explanation in Fig. 2. When similar case data is extracted (S3), the CPU 22 performs a process of matching the time axis of the acquisition timing of the medical image with respect to the plurality of data (S5, S6, S7, S9, S10, S13, S14). The plurality of data to be matched on the time axis may be the similar case data 50 and the target data 40, or may be the extracted plurality of similar case data 50.

In the present embodiment, the user selects one of a method of manually matching the time axis, a method of automatically matching the time axis according to the similarity of medical images, and a method of automatically matching the time axis according to the treatment timing. The instruction to be performed can be input by operating the operation unit 25.

When an instruction to manually match (adjust) the time axis is input (S5: YES), the CPU 22 uses a plurality of data (similar case data 50 and target data 40, or a plurality of data) for which the time axis is to be matched. The medical image of the similar case data 50) of the above is displayed on the display unit 26 together with the time axis (S6). In the example shown in FIG. 4, the medical images 41A and 41B included in the target data 40 and the medical images 51A, 51B, 51C and 51D included in the extracted similar case data 50A are displayed on the display unit 26 together with the time axis T. Has been done. On the time axis T, the timing of taking each medical image is shown. For example, the imaging timing of the medical image 41A is indicated by "A" on the time axis T, and the imaging timing of the medical image 51D is indicated by "d" on the time axis T. Further, in the example shown in FIG. 4, the treatment timing at which the treatment is executed or started is also shown based on the treatment information included in the similar case data 50A. In the example shown in FIG. 4, the timing of starting the medication coincides with the timing of photographing the medical image 51B.

The CPU 22 matches the time axis of the acquisition timing of the medical image between the plurality of data (in the example shown in FIG. 4, between the target data 40 and the similar case data 50A) in response to an instruction input from the user. Further, the CPU 22 changes the imaging timing of each medical image shown on the time axis T according to the result of matching the time axes (S7). The CPU 22 may change the display positions of a plurality of medical images and the like according to the matched time axis together with the imaging timing on the time axis T. For example, the user may input an instruction to match the time axis by inputting an instruction to slide the shooting timing indicated on the time axis T in a direction along the time axis by the operation unit 25. Further, the user can use two medical images whose imaging timings match among the medical images included in each of the plurality of data (in the example shown in FIG. 4, the medical image 41A of the target data 40 and the medical image of the similar case data 50A). You may enter an instruction to specify image 51A). In this case, the CPU 22 matches the time axes of the plurality of data so that the imaging timings of the two designated medical images match.

When an instruction to match the time axis according to the similarity of the medical images is input (S9: YES), the CPU 22 has a plurality of data (similar case data 50 and target data 40, or a plurality of similar case data 50). By comparing the similarity of the medical images included in each of the data among the data and matching the imaging timings of the medical images having the highest similarity, the time axes of the plurality of data are matched (S10). The CPU 22 causes the display unit 26 to display both the medical image included in each of the plurality of data and the time axis T indicating the shooting timing of the medical image as medical information in the state where the time axes are matched in S10 (S11). ). When comparing medical images of a plurality of data, the disease states in the two medical images having the highest similarity are likely to be similar. Therefore, by performing the processing of S10, the user can more appropriately grasp the transition of the disease from the medical information.

When an instruction to match the time axis according to the treatment timing is input (S13: YES), the CPU 22 performs treatment between a plurality of data (similar case data 50 and target data 40, or a plurality of similar case data 50). By matching the timings, the time axes of a plurality of data are matched (S14). The CPU 22 causes the display unit 26 to display both the medical image included in each of the plurality of data and the time axis T indicating the shooting timing of the medical image as medical information in the state where the time axes are matched in S14 (S15). ). By performing the process of S14, the user can compare a plurality of data with reference to the treatment timing, so that the transition of the disease due to the treatment can be appropriately predicted. In addition, when the treatment is not yet executed for the diagnosis target person for which the target data 40 is acquired, the time is based on the timing of the treatment assuming that the diagnosis target person is to be treated in the future. The axes may be aligned. In this case, the user may input the timing of future treatment, or the timing of treatment may be the time when the latest medical image 41 is taken.

Next, the CPU 22 determines whether or not an instruction for displaying the predicted image of the diagnosis target person has been input by the user (S17). The predicted image is an image that is predicted to show the future state of the disease of the person to be diagnosed. When an instruction to display the predicted image is input (S17: YES), the CPU 22 generates a predicted image based on the medical image 41 included in the target data 40 and the medical image 51 included in the similar case data. , Displayed on the display unit 26 (S18).

An example of a method for generating the predicted image 60 will be described with reference to FIG. First, the CPU 22 extracts the medical image 41 used for generating the predicted image as the base image 42 from the medical image 41 included in the target data 40 (see FIGS. 3 and 4) for the diagnosis target person. The base image 42 used to generate the predicted image is preferably a new image as much as possible. Therefore, when the target data 40 includes a plurality of medical images 41, the CPU 22 extracts the medical image 41 having the latest imaging timing as the base image 42. Next, the CPU 22 captures the medical image 51 captured after the same timing as the imaging timing of the base image 42 on the matched time axis among the plurality of medical images 51 included in the similar case data 50 (see FIG. 4). , Extracted as a reference image 52. In the present embodiment, the reference image 52 is extracted in a state where the time axes of the imaging timings of the target data 40 and the similar case data 50 are matched. Therefore, the reference image 52 for generating the predicted image 60 is extracted more appropriately.

In the example shown in FIG. 5, the base image 42 includes the diseased portion 43, and the reference image 52 also includes the diseased portion 53. However, in the reference image 52, the disease progresses as a result of the patient not being treated, and the diseased portion 53 larger than the diseased portion 43 of the base image 42 is reflected.

The CPU 22 generates a predicted image 60 based on the base image 42 and the reference image 52. As an example, in the present embodiment, the CPU 22 generates a diseased portion removal image 45 in which the information of the diseased portion 43 is removed from the basic image 42. Further, the CPU 22 generates a diseased part image 55 in which information other than the diseased part 53 is removed from the reference image 52. The CPU 22 generates a predicted image 60 based on the diseased part removal image 45 and the diseased part image 55 (for example, by performing composition by image processing or the like). As a result, based on the base image 42 and the reference image 52, a predicted image predicted to show the future disease state of the diagnosed subject is appropriately generated.

It is also possible to change the specific generation method of the predicted image 60. For example, the predicted image 60 may be generated by utilizing a mathematical model trained by a machine learning algorithm. In this case, for example, the mathematical model may be pre-trained with data from a plurality of medical images so that the predicted image 60 is output by inputting the base image 42 and the reference image 52. Further, a mathematical model that inputs the base image 42 and outputs the diseased part removal image 45, a mathematical model that inputs the reference image 52 and outputs the diseased part image 55, and the diseased part removal image 45 and the diseased part image 55. At least one of the mathematical models that inputs and outputs the predicted image 60 may be used.

Return to the explanation in Fig. 2. Whether or not the CPU 22 has been input with an instruction to display a progress transition graph showing the transition of the disease progression in each of the plurality of data (similar case data 50 and target data 40, or a plurality of similar case data 50). Is determined (S20). When an instruction to display the progress progress graph is input (S20: YES), the CPU 22 acquires information on the progress of the disease shown in each of the plurality of medical images (S21). The CPU 22 generates a progress transition graph showing the transition of the progress in each data in a state where the time axes of the shooting timings among the plurality of data are matched, and displays the graph on the display unit 26 (S22).

An example of a method for generating a progress transition graph will be described with reference to FIG. In the example shown in FIG. 6, in the case data extraction process (see S3 and FIG. 2), similar case data (black circle graph) of patients who received both surgery and medication, and similar cases of patients who received only medication. Data (black square graph) and similar case data (white circle graph) for patients who were not treated are extracted from each case data group. In addition, the transition of the progress of the test subject is shown by a white square graph.

First, the CPU 22 acquires information on the degree of progression of the disease shown in each of the medical images included in each data. The method for acquiring information on the degree of disease progression can be appropriately selected. For example, the CPU 22 may acquire information on the degree of progression by inputting a medical image into a mathematical model trained by a machine learning algorithm and outputting the degree of progression (for example, the probability of a disease) to the mathematical model. In addition, the CPU 22 inputs a medical image into a mathematical model that outputs the analysis result of the disease, and information indicating the distribution of the degree of influence that the mathematical model has influenced when outputting the analysis result (sometimes referred to as an "attention map"). Information on the degree of disease progression may be obtained based on (there is). Further, the CPU 22 may perform image processing on each medical image and acquire information on the degree of progression based on at least one of the size and color of the diseased portion. The above-mentioned attention map may be displayed on the display unit 26 together with the original medical image.

For each of the plurality of data (target data 40 and three similar case data 50 in the present embodiment), the CPU 22 obtains a progress transition graph showing the transition of the disease progression from the medical image. Create based on the information in. Here, the CPU 22 generates a progress transition graph in a state where the time axes of the shooting timings of the plurality of data are matched. In the example shown in FIG. 6, in S13 of FIG. 2, the time axis of the imaging timing is matched so that the treatment timing in the data of the patient who performed the treatment is matched. Therefore, in the example shown in FIG. 6, the user can easily compare the transition of the degree of progression of the disease according to the presence or absence of treatment and the content of treatment based on the treatment timing.

In the example shown in FIG. 6, among the plurality of medical images 41 included in the target data 40, the imaging timing of the latest medical image 41 is adjusted to the treatment timing of other data as a provisional treatment timing. Therefore, the user can easily predict the transition of the disease when the treatment is started immediately by the progress transition graph. Further, in the example shown in FIG. 6, among the similar case data of the patients who were not treated, the imaging timing of the medical image 51 having the highest degree of similarity to the latest medical image 41 in the target data 40 is the target data. It is adjusted to the shooting timing of the latest medical image 41 in 40.

Further, when the similar case data extracted in S3 includes the above-mentioned heterogeneous data, the CPU 22 displays the heterogeneous data together with the medical image included in the similar case data or separately from the medical image. It is displayed on the unit 26. Therefore, the user can more appropriately predict the transition of the disease of the diagnosis target person by confirming the heterogeneous data included in the similar case data. Further, even when the heterogeneous data regarding the diagnosis target person has not been acquired yet, the user can confirm the heterogeneous data of the patients having similar cases, so that the diagnosis of the diagnosis target person can be performed more appropriately. be able to.

The technology disclosed in the above embodiment is only an example. Therefore, it is possible to modify the techniques exemplified in the above embodiments. For example, the CPU 22 of the medical information processing apparatus 20 may specify the timing (change timing) at which the change in the feature amount of the plurality of medical images included in each data becomes equal to or greater than the threshold value for each of the plurality of data. The CPU 22 may match the time axis by matching the change timings in the data. The change timing may be, for example, the timing when the change in the progress of the disease (progress of deterioration) becomes equal to or higher than the threshold value, or the change timing may be the timing when the change in the progress of healing of the disease becomes equal to or higher than the threshold value. Further, the CPU 22 obtains similar case data by comparing the medical image before or after the change timing with the medical image 41 of the target data 40 among the plurality of medical images 51 included in the case data 50. It may be extracted.

The process of acquiring the target data 40 in S1 of FIG. 2 is an example of the “target data acquisition step”. The process of extracting similar case data in S3 of FIG. 2 is an example of the “similar case data extraction step”. The process of matching the time axis of the shooting timing in S7, S10, and S14 of FIG. 2 is an example of the “time axis matching step”. The process of outputting medical information in S7, S11, S15, S18, and S22 of FIG. 2 is an example of the “medical information output step”. The process of generating the predicted image in S18 of FIG. 2 is an example of the “predicted image generation step”. The process of acquiring the progress information in S21 of FIG. 2 is an example of the “progress acquisition step”. The process of generating the progress transition graph in S22 of FIG. 2 is an example of the “graph generation step”.

10 Server 13 Storage device 20 Medical information processing device 22 CPU
23 Storage device 26 Display unit 40 Target data 41 Medical image 50 Case data 51 Medical image 60 Predicted image

Claims (11)

A medical information processing program executed in a medical information processing device that outputs medical information useful for diagnosis based on case data similar to the data of a person to be diagnosed among a plurality of case data.
Each of the plurality of case data is stored in a database in a state including data of a plurality of medical images taken at different timings and information on the timing of taking each medical image.
When the medical information processing program is executed by the control unit of the medical information processing device,
A target data acquisition step for acquiring at least one medical image data regarding a diagnosis target person and target data including information on the timing of taking the medical image, and a target data acquisition step.
A similar case data extraction step of extracting the case data including a medical image similar to at least one medical image included in the target data from the plurality of case data as similar case data.
A time axis matching step of matching the time axis of the acquisition timing of the medical image between the similar case data, the target data, and at least one of the plurality of similar case data.
A medical information output step that outputs medical information based on the similar case data in a state where the time axes are matched, and
Is executed by the medical information processing apparatus. The medical information processing program according to claim 1.
In the time axis matching step, the control unit is a medical information processing program characterized in that the time axis of the acquisition timing of a medical image among a plurality of data is matched according to an instruction input from a user. The medical image processing apparatus according to claim 1 or 2.
In the time axis matching step, the control unit is characterized in that the time axes are matched by matching the imaging timings of the medical images having the highest degree of similarity among the medical images included in each of the plurality of data. Medical information processing program. The medical information processing program according to any one of claims 1 to 3.
A medical information processing program characterized in that the case data includes treatment information which is information on treatment performed on a patient. The medical information processing program according to claim 4.
In the similar case data extraction step, the control unit is a plurality of case data groups classified in the plurality of case data according to at least one of the presence or absence of treatment indicated by the treatment information and the content of the performed treatment. A medical information processing program characterized by extracting the similar case data from one or a plurality of the case data groups. The medical information processing program according to claim 4 or 5.
The treatment information included in the case data includes information indicating the treatment timing when the treatment is executed or started.
In the time axis matching step, the control unit is a medical information processing program characterized in that the time axis is matched by matching the treatment timing between a plurality of data. The medical image processing program according to any one of claims 1 to 6.
In the similar case data extraction step, when the target data acquired in the target data acquisition step includes a plurality of medical images, the control unit determines the imaging interval of the plurality of medical images in the target data. A medical information processing program characterized by extracting the case data in which a plurality of medical images are similar to each other as the similar case data from the case data including a plurality of medical images in which the difference between the two is equal to or less than a threshold. .. The medical information processing program according to any one of claims 1 to 7.
In the medical information output step, the control unit sets both the medical image included in each of the plurality of data and the time axis indicating the imaging timing of the medical image in a state where the time axes are matched. A medical information processing program characterized by displaying medical information on the display unit. The medical information processing program according to any one of claims 1 to 8.
The control unit
Of the medical images included in the case data extracted in the similar case data extraction step, the medical images included in the target data are captured after the same timing as the imaging timing of the medical images included in the target data on the matched time axis. A predictive image generation step of generating a predictive image of the diagnosis target person based on the medical image and the medical image included in the target data is further executed.
In the medical information output step, the control unit displays the predicted image generated in a state where the time axes are matched on the display unit as the medical information. The medical information processing program according to any one of claims 1 to 9.
The control unit
The progress acquisition step for acquiring information on the progress of the disease shown in each of multiple medical images,
Graph generation that generates a progress transition graph showing the transition of the progress in each of the similar case data and the target data, or a plurality of the similar case data, in a state where the time axes of the plurality of data are matched. Steps and
Further run,
In the medical information output step, the medical information processing program is characterized in that the progress transition graph generated in a state where the time axes are matched is displayed on the display unit as the medical information. A medical information processing device that outputs medical information useful for diagnosis based on case data similar to the data of a person to be diagnosed among a plurality of case data.
Each of the plurality of case data is stored in a database in a state including data of a plurality of medical images taken at different timings and information on the timing of taking each medical image.
The control unit of the medical information processing device
A target data acquisition step for acquiring at least one medical image data regarding a diagnosis target person and target data including information on the timing of taking the medical image, and a target data acquisition step.
A similar case data extraction step of extracting the case data including a medical image similar to at least one medical image included in the target data from the plurality of case data as similar case data.
A time axis matching step of matching the time axis of the acquisition timing of the medical image between the similar case data, the target data, and at least one of the plurality of similar case data.
A medical information output step that outputs medical information based on the similar case data in a state where the time axes are matched, and
A medical information processing device characterized by executing.

Images