JP2018130408A - Control method of information terminal - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve an information terminal for searching for a similar medical image similar to a medical image to be read.SOLUTION: A display screen displays a plurality of similar medical images by dividing them every image observation name of a plurality of first division layers one layer below image observation names in a highest layer. In the case that selection of any of the image observation names in the plurality of first division layers is detected, when image observation names in a selected first division layers are in a first layer set to additional information of similar medical images, the display screen including all disease names set to the image observation names in the first layer associated with image observation names in the selected first division layer is displayed to a display; when a layer one layer below the image observation names in the selected first division layer is image observation names in a second division layer higher than the first layer, the display screen displaying a plurality of similar medical images associated with the image observation names in the selected first division layer by being divided by the image observation names in the second division layer is displayed to the display.SELECTED DRAWING: Figure 25

Description

  The present invention relates to an information terminal control method for searching for a similar medical image similar to a medical image to be interpreted.

  In recent years, medical imaging apparatuses such as CT (Computed Tomography) and MRI (Magnetic Resonance Imaging) have been developed and spread. These CT and MRI have made it possible to acquire high-capacity digitalized high-definition medical images. Further, medical images read by doctors are being sequentially accumulated in PACS (Picture Archiving and Communication Systems) together with interpretation reports. Here, as disclosed in, for example, Patent Documents 1 and 2, a technique for searching past medical images similar to medical images to be interpreted from past cases accumulated in the PACS in order to be used as a reference for new interpretation. Has begun to be developed.

  Patent Document 1 discloses a technique for searching a representative case image from a disease-specific case DB and displaying a disease-specific case image as a search result.

  Patent Document 2 discloses a technique for performing additional learning so that specific case data (such as misdiagnosis, interpretation by a well-known doctor, and the number of detailed display requests are large) are searched with high frequency.

  Patent Document 3 discloses a technology that allows a medical information management system and a case search system to be linked and used without being aware of the difference between the two systems.

JP 2008-257292 A JP 2007-279942 A International Publication No. 2015/155807

Nemoto, Shimizu, Sakakibara, Obata, Nawano, “Improvement of mass shadow discrimination accuracy on mammograms by feature selection from many features and proposal of fast feature selection method”, IEICE Transactions D- II, Vol. J88-D-II, no. 2, pp. 416-426, February 2005

  Patent Document 3 discloses a technique for displaying a large number of thumbnail images of similar cases (for example, 20 images on one screen in FIG. 6). However, Patent Documents 1 to 3 show a number of similar case search results. There has been no disclosure of a technique for selecting similar cases to be used as reference for disease differentiation from a number of similar case images by separately classifying and displaying them.

  One aspect of the present disclosure is a method for controlling an information terminal that is connected to a case retrieval system that retrieves a medical image with reference to a medical image database, and includes (a) a medical image to be interpreted and a predetermined A plurality of similar medical images having the above similarities are received from the case search system, each of the plurality of similar medical images has additional information, and the additional information of the similar medical image includes a disease name and a first hierarchy The image finding name of the first layer is associated with at least one image finding name of a layer higher than the first layer, and is higher than the first layer. The upper level image finding name includes the highest level image finding name, and the uppermost level image finding name is associated with all the first level image finding names of the plurality of similar medical images. Being (B) a display screen including the plurality of similar medical images is displayed on the display, and the display screen includes a plurality of images that are one layer lower than the highest level image finding name. The plurality of similar medical images are displayed separately for each image finding name of the first classification hierarchy, and (c) selection of any of the image finding names of the plurality of first classification hierarchy is detected. (I) when the image finding name of the selected first division hierarchy is the first hierarchy set in the additional information of the similar medical image, the selected first division hierarchy A display screen including all disease names set in the additional information including the image finding name of the first hierarchy associated with the image finding name is displayed on the display; (ii) the selected first The hierarchy one level below the image finding name of the classification hierarchy is In the case of the image finding name of the second classification hierarchy that is a hierarchy above the first hierarchy, a plurality of similar medical images associated with the image finding name of the selected first classification hierarchy are A display screen that is divided and displayed by the image finding names of the second division hierarchy is displayed on the display.

  According to the above aspect, by sorting and displaying a large number of similar case search results according to findings, it is possible to support for selecting and distinguishing diseases from a large number of similar case images.

1 is an overall configuration diagram of a hospital information system to which an information terminal according to Embodiment 1 is applied. It is a block diagram which shows the structure of an information terminal, a medical information management system, and a case search system. It is an external view of two displays. It is a figure which shows the structural example of the mounting form of an information terminal. It is a figure which shows an example of the basic screen displayed on a display immediately after starting a similar case search application with an information terminal. It is a figure which shows an example of the basic screen displayed on a display immediately after starting a similar case search application with an information terminal. It is the figure which extracted and showed the display area of a certain similar case displayed on a case display area. It is an enlarged view of a disease name list display area. It is a figure which shows a basic screen when a similar case is narrowed down by "mycosis." It is a figure which shows the data structure of patient information. It is a figure which shows the data structure of the test | inspection information registered into the patient information shown in FIG. It is a figure which shows the data structure of a medical image database. It is a figure which shows the data structure of a diagnostic report. It is a figure which shows the data structure of similar case data. It is the figure which showed typically the region of interest set to the slice image. It is a figure which shows the data structure of a disease system. FIG. 5 is a sequence diagram showing processing after an information terminal acquires a diagnosis target case from a medical information management system, requests a case search system for a similar case search, and the case search system receives a request for similar case search. is there. It is a screen figure of an inspection list. It is a screen figure of an inspection list after an inspection is selected. It is the figure which showed the slice image displayed on a medical image viewer, when a series is selected by the user. It is a sequence diagram which shows the process which returns a similar case search result to an information terminal, and displays the initial screen of a search result, after a case search system receives the request | requirement of a similar case search. It is a figure which shows an example of an image finding system. It is a flowchart of the process which produces | generates the initial screen of a search result. It is the figure which showed the number of cases for every image finding of all the cases which the information terminal acquired as a search result together with the image finding system of FIG. It is the figure shown about the initial screen which displays a search result in an information terminal. It is another example of the figure which showed the number of cases for every image finding of all the cases which the information terminal acquired as a search result together with the image finding system of FIG. It is another example of the figure shown about the initial screen which displays a search result in an information terminal. It is the figure shown about the table information produced | generated as a disease name list. It is an example of a display of a disease name list. It is an example of a display of a disease name list. It is an example of a display of a disease name list. It is the figure shown about the screen which narrowed down and displayed the search result by the image finding in the information terminal. It is the figure shown about the screen which narrowed down and displayed the search result by the image finding in the information terminal. It is the figure shown about the screen which highlighted the disease name list in the information terminal. It is the figure shown about the screen where a disease name list is not initially displayed in an information terminal. It is the figure which showed the screen where the disease name list was displayed on the screen in which the disease name list is not initially displayed in the information terminal. It is the figure shown about the screen which provided the button which displays a disease name list, when a disease name list is not initially displayed in an information terminal. It is the figure which showed the example which contains multiple types of lesions in the region of interest. FIG. 10 is a diagram showing a data configuration of similar case data in the second embodiment. 10 is a flowchart of processing for generating an initial screen of search results in the second embodiment. 10 is a display example of a display icon / selection button for an image finding name in the second embodiment. 10 is a display example of a display icon / selection button for an image finding name in the second embodiment. FIG. 10 is an operation example when selecting a display icon / selection button for an image finding name in Embodiment 2. FIG. FIG. 10 is an operation example when selecting a display icon / selection button for an image finding name in Embodiment 2. FIG. FIG. 10 is an explanatory diagram of a data configuration of similar case data in the third embodiment. 10 is an explanatory diagram illustrating a display example of image findings and clinical information in Embodiment 3. FIG.

(Knowledge that is the basis of the present invention)
First, the knowledge that forms the basis of the present invention will be described. Using the search result display screen disclosed in Patent Document 3, many similar cases of the interpretation target cases were presented, and an experiment was conducted to verify the influence on the image diagnosis result.

  The experimental method was as follows. The subjects were five radiologists (hereinafter referred to as “specialists”) and four first-stage or late-stage trainers (radiologist) (hereinafter referred to as “residents”). First, the subject performs image diagnosis on the diagnosis target image (total of 3 questions) without reference information.

  As a result of image diagnosis, an image finding and a diagnosis disease name are described in the interpretation report. Thereafter, using the search result display screen disclosed in Patent Document 3, a large number of pairs of similar cases and definite diagnosis disease names of the interpretation target cases are presented to the subject. The test subject can correct the interpretation report with reference to a large number of sets of similar case images and definite diagnosis disease names. In addition, two radiology specialists different from the subjects listed for each problem the image findings to be listed (hereinafter “appropriate findings”) and the diseases to be listed as diagnostic imaging disease names (hereinafter “appropriate diseases”). .

  In the experiment, the following results were obtained. After referring to similar cases, the image findings in the interpretation report did not change, and 0.7 diagnostic disease names were added per question for specialists and 1.3 for the residents. From this result, it can be seen that there is an effect of recalling a diagnosis disease name as an effect of similar cases. The proportion of additional diseases that were “appropriate” was 82% for specialists and 44% for residents. From this result, it is presumed that, from among a large number of displayed similar case thumbnail images, the resident can not select the disease additionally described in the interpretation report as the diagnosis of the diagnosis target case as much as the specialist. Furthermore, the ratio of “appropriate findings” among the image findings described in the interpretation report was 64% for specialists and 35% for residents. The correlation coefficient between the “appropriate finding” rate and the “appropriate finding” rate of additional diseases was 0.62 for each problem and subject. From this, it is considered that there is a correlation between the ability to read findings and the ability to select differential diseases from a number of similar case images. Through such experiments and subsequent considerations, the present inventors have come up with the present invention that supports the selection of differential diseases from a number of similar case images by classifying and displaying a number of similar case search results according to findings.

  One aspect of the present disclosure is a method for controlling an information terminal that is connected to a case retrieval system that retrieves a medical image with reference to a medical image database, and includes (a) a medical image to be interpreted and a predetermined A plurality of similar medical images having the above similarities are received from the case search system, each of the plurality of similar medical images has additional information, and the additional information of the similar medical image includes a disease name and a first hierarchy The image finding name of the first layer is associated with at least one image finding name of a layer higher than the first layer, and is higher than the first layer. The upper level image finding name includes the highest level image finding name, and the uppermost level image finding name is associated with all the first level image finding names of the plurality of similar medical images. Being (B) a display screen including the plurality of similar medical images is displayed on the display, and the display screen includes a plurality of images that are one layer lower than the highest level image finding name. The plurality of similar medical images are displayed separately for each image finding name of the first classification hierarchy, and (c) selection of any of the image finding names of the plurality of first classification hierarchy is detected. (I) when the image finding name of the selected first division hierarchy is the first hierarchy set in the additional information of the similar medical image, the selected first division hierarchy A display screen including all disease names set in the additional information including the image finding name of the first hierarchy associated with the image finding name is displayed on the display; (ii) the selected first The hierarchy one level below the image finding name of the classification hierarchy is In the case of the image finding name of the second classification hierarchy that is a hierarchy above the first hierarchy, a plurality of similar medical images associated with the image finding name of the selected first classification hierarchy are A display screen that is divided and displayed by the image finding names of the second division hierarchy is displayed on the display.

  According to this aspect, a plurality of similar medical images received from a case search system that searches a medical image with reference to a medical image database are classified into a plurality of first classifications that are one layer below the highest level image finding name. The plurality of similar medical images are classified and displayed for each image finding name in the hierarchy. In this way, by segmenting based on image finding names that are associated in advance, even if the user is not an expert in image diagnosis, compared to a method in which similar medical images are simply displayed in order of similarity, etc. In addition, it is possible to expect an effect that makes it easy to grasp the image findings of similar medical images.

  In the above aspect, for example, in the step (b), the image finding names of the plurality of first division layers are displayed so as to be selectable, and in the step (c), the plurality of first divisions are displayed. Selection of any of the image finding names in the hierarchy may be detected. As a result, the image finding name serves as both a display and a selection button, and an effect of reducing the burden on the user's operation and line-of-sight movement can be expected.

  In the above aspect, for example, in the step (b), the display screen includes a display for accepting a request for displaying a disease name of the user, and (d) when detecting a request for displaying the disease name of the user, A display screen including all the disease names set in the image finding names of the first hierarchy associated with all the image finding names displayed on the display screen when detecting the user's disease name display request May be displayed on the display. As a result, when similar medical images are narrowed down to appropriate image findings, even if the image findings displayed on the display screen do not reach the first layer image findings, When you want to see the list, you can check the display screen that includes the disease name.

  In the above aspect, for example, in the step (b), a predetermined number of medical images selected from the plurality of medical images are displayed on the display screen, and the plurality of first classification layers are displayed. A plurality of medical images having a predetermined variance or more in the image feature space may be selected from all the medical images corresponding to each of the image finding names and displayed on the display screen. This makes it easy for the user to understand the features of the image findings by showing various variations of the images in the medical image group associated with the same image finding name.

  In the above aspect, for example, in the step (b), the plurality of medical images are in the order of similarity in an image feature space with the medical image to be interpreted, and a plurality of first classification layers The image findings may be displayed separately for each image finding name. As a result, an image having a high similarity to the search query image is displayed from the group of medical images associated with the same image finding name, so that the image finding is selected for easy discrimination from other image findings. It becomes easy.

  In the above aspect, for example, in the step (b), at least two of the plurality of similar medical images are selected as the first image finding name from the plurality of image finding names of the first classification hierarchy. When a predetermined number or more of similar medical images associated with the combination are included, the combination of the image finding names of the plurality of first classification layers may be displayed separately. By confirming a list of similar medical images associated with a plurality of image finding names, the user can easily grasp variations of similar medical images associated with a plurality of image findings.

  In the above aspect, for example, at least one or more pieces of clinical information is set in the additional information of the similar medical image, and in the step (b), the first medical image is added to the plurality of similar medical images. When a predetermined number or more of similar medical images having additional information including a combination of the image findings of the hierarchy and the clinical information are included, the combinations of the image findings of the first hierarchy and the clinical information are also displayed separately. You may be allowed to. By checking a list of similar medical images in which the combination of image findings and clinical information matches, it becomes easier for the user to grasp similar medical images close to the disease to be diagnosed.

  In the above aspect, for example, at least one disease progress information is set in the additional information of the similar medical image. In the step (b), the plurality of similar medical images are When a predetermined number or more of similar medical images having additional information having a combination of the first layer image findings and the disease progress information are included, the first layer image findings and the progress information Combinations may also be displayed separately. By confirming a list of similar medical images in which the combination of the image findings and the degree of progress information matches, it becomes easier for the user to grasp similar medical images close to the disease to be diagnosed.

  In the above aspect, for example, (d) in step (ii) of the step (c), a display screen that is divided and displayed by the image finding names of the second divided hierarchy is displayed on the display. Thereafter, when selection of the image finding name of the second classification hierarchy is detected, (iii) the image finding name of the selected second classification hierarchy is set in the additional information of the similar medical image In the case of the first hierarchy, a display screen including all disease names set in the image finding names of the first hierarchy associated with the selected image finding names of the second divided hierarchy (Ii) a layer one level lower than the image finding name of the selected second classification hierarchy is an image finding name of a third classification hierarchy that is higher than the first hierarchy. The image location of the selected second segment hierarchy A display screen in which a plurality of similar medical images associated with the name of the view are classified by the image finding name of the third classification hierarchy may be displayed on the display. In this way, the similar medical images can be narrowed down to the detailed image findings even by a less experienced user by sequentially displaying the image findings hierarchy and sorting and displaying the similar medical images by the detailed image findings that are the lower layers. Can be confirmed.

(Embodiment 1)
Hereinafter, an embodiment of the present disclosure will be described with reference to the drawings. In each drawing, the same numerals are used about the same component.

(overall structure)
FIG. 1 is an overall configuration diagram of a hospital information system to which an information terminal according to the present embodiment is applied. As shown in FIG. 1, the hospital information system includes an information terminal 100, a medical information management system 200, and a case search system 300. The medical information management system 200 and the case search system 300 are not necessarily arranged in the hospital, and may be software that operates on a data center outside the hospital, a private cloud server, a public cloud server, or the like. .

  As the information terminal 100, an information terminal such as a personal computer or a tablet terminal is employed. As the medical information management system 200, a PACS (Picture Archiving and Communication Systems), an electronic medical record system, or the like is adopted.

  FIG. 2 is a block diagram showing configurations of the information terminal 100, the medical information management system 200, and the case search system 300. As illustrated in FIG. 2, the information terminal 100 includes a display 101, an operation unit 102, an input control unit 103, a display control unit 104, an ROI management unit 105, a disease name list management unit 106, and a communication control unit 107.

  The display 101 is composed of a liquid crystal monitor, for example, and displays a medical image and medical chart information on which the user performs image diagnosis, a report input screen for entering the result of image diagnosis, and the like. Although at least one display 101 is required, usually two to three are used for image diagnosis. In this embodiment, an example in which two units are used is shown as a first display 101a and a second display 101b, respectively. The display screen of the first display 101a is called a first display screen, and the display screen of the second display 101b is called a second display screen. The state of the two-unit configuration is shown in FIG. In FIG. 3, the screen of the medical image viewer is displayed on the first display 101a, and the screen of the case search system 300 is displayed on the second display 101b.

  The operation unit 102 includes, for example, a keyboard and a mouse, and can input various information to the information terminal 100 such as selection of display contents in medical images and medical chart information displayed on a display, and input of image diagnosis results.

  When the input control unit 103 recognizes a user operation input to the operation unit 102, the input control unit 103 interprets the content of the operation and notifies the other components of the operation content. For example, the input control unit 103 receives the mouse coordinates from the operation unit 102 and compares them with the coordinates of the GUI generated by the display control unit 104 to determine the state in which the button object in the GUI is selected. Notify the operation contents to the component.

  The display control unit 104 generates a GUI (Graphical User Interface) of the information terminal 100 and causes the display 101 to display the GUI.

  The OI management unit 105 manages a region of interest (Region Of Interest) when a similar case search is performed.

  The name list management unit 106 manages a disease name list of similar case thumbnails displayed in the similar case thumbnail display area 710.

  The communication control unit 107 controls communication between the information terminal 100 and the medical information management system 200 and communication between the information terminal 100 and the case search system 300. In addition, the communication control unit 107 receives various data transmission requests from other blocks, transmits them to the medical information management system 200 or the case search system 300, and is also transmitted from the medical information management system 200 or the case search system 300. Data is received and passed to the corresponding block.

  As shown in FIG. 2, the medical information management system 200 includes a patient information storage unit 201, a patient information management unit 202, a medical image data storage unit 203, a medical image data management unit 204, a diagnostic report management unit 205, and a communication control unit 206. Is provided.

  The patient information accumulating unit 201 accumulates clinical information such as a patient's sex, age, and past history, and examination information such as a blood test.

  The patient information management unit 202 manages input / update / output of information stored in the patient information storage unit 201. The medical image data storage unit 203 stores patient examination images.

  The medical image data management unit 204 displays a patient examination image.

  The diagnosis report management unit 205 manages input and display of reports that are patient diagnosis results.

  The communication control unit 206 receives various data transmission requests from other blocks and transmits them to the information terminal 100 or the case search system 300, and receives data transmitted from the information terminal 100 or the case search system 300, Pass to the appropriate block.

  As shown in FIG. 2, the case search system 300 includes a similar case data storage unit 301, an image feature extraction unit 302, a similar case search unit 303, and an image finding system storage unit 304.

  The similar case data storage unit 301 assigns to each of the image features extracted from each of a plurality of cases selected as target data for similar case search among cases managed in advance by the medical information management system 200. The recorded image findings and the respective thumbnail images are accumulated.

  The image feature extraction unit 302 includes the slice image ID (slice ID) and ROI information transmitted from the communication control unit 110 of the information terminal 100, and the slices stored in the medical image data storage unit 203 of the medical information management system 200. Image features are extracted based on the medical image data corresponding to the ID.

  The similar case search unit 303 generates a similar case search result by comparing the image features extracted by the image feature extraction unit 302 with the image features of a plurality of similar cases stored in the similar case data storage unit 301, respectively. To do.

  The image finding system storage unit 304 stores a tree structure system of predefined image findings. The image findings assigned to the cases accumulated in the similar case data accumulation unit 301 are assigned from this system. This system will be explained later.

  The communication control unit 305 receives various data transmission requests from other blocks, transmits them to the information terminal 100 or the medical information management system 200, and receives data transmitted from the information terminal 100 or the medical information management system 200. And pass it to the appropriate block.

  FIG. 4 is a diagram illustrating a configuration example of an implementation form of the information terminal 100. As shown in FIG. 4, the information terminal 100 includes an application 501, an OS (Operating System) 502, a memory 503, and other hardware not shown.

  The application 501 is application software for causing a personal computer or a tablet terminal to function as the information terminal 100, and is executed by the processor of the information terminal 100. The information terminal 100 may implement the application 501 by reading the application 501 from a computer-readable recording medium, or may implement the application 501 by downloading from a network. The OS 502 is basic software of the portable information terminal, and is executed by the processor of the information terminal 100. The memory 503 is configured by a storage device such as a RAM or a ROM provided in the information terminal 100, and stores a data group included in the application 501. When the processor of the information terminal 100 executes the application 501, the functions of the input control unit 103, the display control unit 104, the ROI management unit 105, the disease name list management unit 106, and the communication control unit 107 shown in FIG. 2 are realized.

  However, in the present embodiment, the information terminal 100 may be implemented solely by the application 501, may be implemented by the application 501 and the OS 502, or may be implemented by the application 501, the OS 502, and the memory 503. Alternatively, the application 501, the OS 502, the memory 503, and other hardware (not shown) may be implemented. In any of the mounting forms, it is possible to realize the information terminal 100 of the present embodiment.

  5 and 6 are diagrams showing screen configurations immediately after the similar case search is started on the information terminal 100. FIG. 5 shows the screen of the display 101a, and FIG. 6 shows the screen of the display 101b.

  As shown in FIG. 5, the basic screen of the information terminal 100 displayed on the display 101 a includes a medical image viewer 610. A medical image is usually recorded in a DICOM (Digital Imaging and Communication in Medicine) format, and a medical image viewer 610 is a viewer that can handle DICOM. The medical image handled in the present embodiment is a chest CT image composed of a number of tomographic images (also referred to as slice images). The chest CT image displayed on the medical image viewer can be sliced by operating a mouse or a keyboard. Instead of the chest CT image, an MRI (Magnetic Resonance Imaging) image or a simple X-ray image may be used.

  In the medical image viewer 610, an image of a case from which a similar case search is started is displayed. Of the chest CT image, an image at a slice position (ie, a search query image) at which a radiogram interpreter has started a similar case search is initially displayed. Note that a region of interest (ROI: Region Of Interest) on which a similar case search is performed may be superimposed on the image.

  As shown in FIG. 6, the basic screen of the information terminal 100 displayed on the display 101b is a similar case thumbnail display area 710, a case number display area 712, a disease name narrowing condition display area 713, a diagnosis target case display area 720, and a disease name list. A display area 730 is provided.

  The similar case thumbnail display area 710 is an area where thumbnail images of cases similar to the search query are displayed according to image findings. In the example of FIG. 6, similar cases belonging to the respective image findings of these 711 a to c are displayed in the horizontal direction below the display icons and selection buttons 711 a to c of the image findings names (this “ The meaning of “belongs” will be explained later together with the image observation system). When the number of cases belonging to the image findings is larger than the number of displays according to the image findings (5 in the example of FIG. 6), the horizontal scroll bars 714a to 714c can scroll in the horizontal direction to view other cases. it can. In the example of FIG. 6, three image findings 711a to 711c are displayed. However, if a case corresponding to four or more image findings is obtained as a search result, a vertical scroll bar 715 is used to Scroll in the direction to view similar cases corresponding to other image findings.

  The number-of-cases display area 712 is an area indicating the number of cases after being narrowed down by the functions of the image finding display icon / selection button 711 and the disease name list 730 among similar cases acquired as a search result.

  The disease name narrowing down condition display area 713 is an area for displaying conditions narrowed down by the function of the disease name list 730.

  Further, the display area of each thumbnail case is configured by a definite diagnosis disease name display area 716 and a distance display area 717 as shown in FIG.

  Here, a little supplement will be made regarding the conversion of image resolution and pixel values for thumbnail images. Although the resolution of the original CT image is 512 × 512 pixels, the thumbnail image may have a lower resolution due to the limitation of the size of the display area. Furthermore, the range of pixel values (CT values) of the original CT image is −1000 to +1000 (unit HU: Hounsfield Unit), and cannot be displayed on a normal 8-bit gradation display as it is. Even if it can be displayed, within 2000 gradations, emphysema area (CT value: -1000HU), normal tissue of lung field (CT value: about -900HU), ground glass area (CT value: -800HU), soft tissue It is difficult for humans to distinguish (CT value: −100 to −50 HU), water (CT value: 0 HU), and bone (CT value: 1000 HU). Therefore, normally, a window level and a window width are separately provided, and conversion into 8-bit luminance for display on the display is performed.

  A plurality of (15 in the example of FIG. 5) similar cases are displayed in the similar case thumbnail display area 710, and image resolution and pixel value conversion from the original CT image (DICOM format) is performed each time the thumbnail is displayed. In order to improve the display speed, it is desirable to create a thumbnail image by performing resolution conversion and pixel value conversion from an original DICOM image in advance.

  Note that the displayed thumbnail images are cases where the distance between feature vectors is equal to or less than a predetermined threshold. Here, the Euclidean distance is used as the distance. Note that another distance scale such as a city distance may be used. Further, the similarity may be used instead of the distance between feature vectors. A cosine similarity or the like can be used as the similarity.

  Furthermore, cases displayed in the similar case thumbnail display area 710 can be narrowed down by operating the disease name list display area 730. The currently displayed case narrowing conditions are displayed in the display condition display area 713. In the state immediately after the similar case search in FIG. 6, nothing has been narrowed down, so “all diseases” is displayed in the display condition display area 713.

  In the disease name list display area 730, the definite diagnosis disease names of all similar cases displayed in the similar case thumbnail display area 710 are displayed. Since cases after a definitive diagnosis are used as cases to be searched for similar cases, a definitive diagnosis disease name is given to each case in advance. FIG. 8 shows an enlarged view of the disease name list. In FIG. 8, the disease names are classified into major classification disease names (731, 734, 737, 741, 744) and detailed classification disease names (732, 733, 735, 736, 738, 739, 740, 742, 743, 745). They are displayed separately. In addition, the number of cases of the disease name is displayed beside the major classification disease name and the detailed classification disease name. The user can limit the cases displayed in the similar case thumbnail display area 710 by selecting an arbitrary row of the major classification disease name or the detailed classification disease name in the disease name list. As shown in FIG. 6, in the state immediately after the similar case search, 62 cases including various diseases are to be displayed. However, when the row 731 in FIG. 8 is clicked with a mouse, the major classification disease names as shown in FIG. Is displayed only for cases with mycosis (14 cases in total). At this time, the display condition display area 714 is changed to a display of “mycosis” in FIG. 9 so that the narrowing-down conditions of the currently displayed case can be understood. With this function, for example, by displaying a plurality of similar cases of the disease for a disease name recalled by a doctor who performs image diagnosis, it becomes easy to confirm consistency with the disease name recalled by the case to be diagnosed.

  FIG. 10 is a diagram showing a configuration of patient information 1000. As shown in FIG. The patient information 1000 is managed for each patient by the patient information storage unit 201 in the medical information management system 200. As shown in FIG. 10, the patient information 1000 includes a patient ID 1100, a name 1200, an age 1300, a gender 1400, a past history 1500, a family history 1600, a main complaint 1700, examination information 1800, and a definitive diagnosis 1900. Patient ID 1100 is an identification number unique to the patient. Name 1200, age 1300, sex 1400, medical history 1500, family history 1600, and chief complaint 1700 are the name, age, sex, medical history, family history, and chief complaint of the patient with patient ID 1100, respectively. Moreover, the test | inspection information 1800 represents the information regarding the some test | inspection which the said patient received in the past, as shown in FIG. Each inspection includes an inspection ID 1810, an inspection date 1820, an inspection type 1830, and an inspection result 1840. The inspection ID 1810 is an identification number unique to the inspection. The inspection date 1820 is the date and time when the inspection is performed. The inspection type 1820 is the type of the inspection. For example, there are blood tests, respiratory function tests, endoscopy, simple X-ray imaging, CT imaging, and the like. If it is a blood test, the test result 1840 includes the values of various indexes such as the white blood cell count, LDH, and GPT, and judgments made by the doctor based on the values. In the case of image inspection such as simple X-ray imaging and CT imaging, pointer information to a captured image itself and pointer information to a report of an image diagnosis result are also included. The image itself is stored in the medical image data storage unit 203 in the medical information management system 200 in the DICOM format.

  When the examination type 1830 is an image examination such as simple X-ray, CT, MRI, and PET, the medical image data 2000 is accumulated in the medical image data accumulation unit 203 in the medical information management system 200 as shown in FIG. . The medical image data 2000 has an examination ID 1810 (common with the examination ID of FIG. 11) and a series ID. Since a plurality of types of imaging (for example, simple CT and contrast CT) may be performed in one examination, a plurality of series IDs 2100 may correspond to one examination ID. Further, in the case of CT examination, MRI examination, and the like, a plurality of slice images are acquired by one imaging, and therefore a plurality of slice IDs correspond to one series ID.

  When the examination type 1830 is an image examination such as simple X-ray, CT, MRI, and PET, a diagnostic report 3000 as shown in FIG. 13 is accumulated in the diagnostic report management unit 205 in the medical information management system 200. The diagnosis report 3000 includes an examination ID 1810 (common with the examination ID of FIG. 11), a finding 3100, and a diagnosis 3200.

  FIG. 14 is a diagram illustrating an example of a data configuration of the similar case data 4000. As illustrated in FIG. The similar case data 4000 is data that is referred to when searching for cases similar to the diagnosis target case, and is stored in the similar case data storage unit 301 of the case search system 300. As shown in FIG. 14, the similar case data 4000 includes disease IDs 4400 corresponding to the number of similar case IDs 4100, the number of regions of interest 4200, the number of confirmed diagnoses 4300, and the number of confirmed diagnoses 4300.

  The similar case ID 4100 is an identifier of the similar case data 4000. In the example of FIG. 14, the similar case ID 4100 is composed of a symbol string composed of “SIM” followed by a number.

  The number of regions of interest 4200 is the number of regions of interest set in advance for the case indicated by the similar case ID. The similar case data 4000 includes information of the region of interest ID 4210, the slice ID 4220, the image finding ID 4230, the region of interest coordinates 4240, the image feature data 4250, and the thumbnail image data 4260 as many as the number of regions of interest 4200. The region of interest ID 4210 is the ID of the region of interest assigned to the similar case indicated by the similar case ID 4100. Therefore, the region of interest ID 4210 is expressed by giving a serial number after the symbol string indicated by the similar case ID 4100. The slice ID 4220 is an identifier of the slice image in which the region of interest is set, and is the same as the slice ID 2200 shown in FIG. The image finding ID 4230 is an image finding ID set from an image finding system described later. The region of interest coordinates 4240 is information indicating the position of the region of interest set in the slice image. FIG. 15 is a diagram schematically showing a region of interest set in a slice image. In the example of FIG. 15, the region of interest is set in a rectangular shape. Accordingly, the region-of-interest coordinates 4240 is composed of four values, the coordinates (xl, yt) of the upper left vertex of the region of interest and the coordinates (xr, yb) of the lower right vertex. Of course, the region of interest may have a shape other than a rectangle. In this case, a parameter that can uniquely identify the region is adopted as the region of interest coordinates 4240. For example, if the region of interest is a circle, the center coordinates and radius of the circle are adopted as the region of interest coordinates 4240. The image feature data 4250 is a feature value of a predetermined dimension (N dimension here) extracted from the region of interest defined by the region of interest coordinates 4240. The thumbnail image data 4260 is image data of a thumbnail image generated for displaying in the case display area 710 based on the slice image in the DICOM format specified by the slice ID. In the present disclosure, a plurality of regions of interest can be set across different slice images. Therefore, there are a plurality of thumbnail image candidates for the similar case ID 4100. Here, it is assumed that a thumbnail image is created and held for each slice image to which a region of interest is assigned. In the thumbnail image data 4260, for example, pixel values of thumbnail images are arranged in raster scanning order from the upper left vertex to the lower right vertex of the thumbnail image. As described above, the DICOM image obtained by the CT examination is an 11-bit (pixel value: −1000 to +1000) image of 512 × 512 pixels. Therefore, in the present embodiment, in order to speed up the display of thumbnail images, low resolution processing and gradation conversion processing are performed on the DICOM image that is the source of the thumbnail image, and the thumbnail has an 8-bit pixel value. An image is created in advance and registered in similar case data 4000. The thumbnail image may be created by, for example, the medical information management system 200 and transmitted to the case search system 300, or the case search system 300 may acquire a DICOM image from the medical information management system 200. You may create it.

  The definite diagnosis number 4300 is the number of disease names given as definitive diagnosis. In this example, since two disease names are given to the case indicated by the similar case ID 4100, “2” is registered in the number of confirmed diagnoses 4300.

  The disease ID 4400 stores disease IDs defined in the disease system 5000 (FIG. 16) for the number of confirmed diagnoses 4300. In the example of FIG. 14, since the disease name indicated by “DIS012_007” and the disease name indicated by “DIS015_019” are given to the case indicated by the similar case ID, “DIS012_007” and “DIS012_019” are registered in the disease ID 4400. Has been.

  Next, the flow from when the information terminal 100 starts the interpretation work to the start of the similar case search in cooperation with the medical image management system 200 and the case search system 300 will be described.

  FIG. 17 shows that after the information terminal 100 acquires the interpretation target case from the medical information management system 200, the case search system 300 requests the similar case search, and the case search system 300 receives the similar case search request. It is a sequence diagram which shows the flow until. First, the information terminal 100 accepts an interpretation target examination list display request of a user (a doctor who performs interpretation) through the operation unit 102, and interprets the interpretation to the communication control unit 206 of the medical image management system 200 through the input control unit 103 and the communication control unit 110. A target examination list display request is transmitted (step S510). The communication control unit 206 transmits an interpretation target examination list display request to the patient information management unit 202. The patient information management unit 202 lists examinations that have undergone image examination and have not yet been interpreted, and transmits the examinations to the communication control unit 110 of the information terminal 100 as an interpretation target examination list through the communication control unit 206 (step S520). The display control unit 104 of the information terminal 100 receives the interpretation target examination list and displays the interpretation target examination list (step S530). For example, a screen as shown in FIG. 18 is displayed. The screen in FIG. 18 includes an area 800 that displays an examination list that has not yet been interpreted, and an area 810 that displays series information included in the examination ID. Here, since the inspection (corresponding to the row) in the inspection list 800 is not selected, nothing is displayed in the area 810.

  The user selects an examination to be interpreted from the list. When the selection is performed, the communication control unit 110 transmits a display request for the inspection image of the target inspection ID to the medical information management system 200 (step S540). The medical information management system 200 acquires from the medical image data storage unit 203 all series images included in the examination ID designated by the patient information management unit 202 (see FIG. 12 for the relationship between examinations and series), and the communication control unit 206. Then, the information is transmitted to the information terminal 100 (step S550). The display control unit 104 of the information terminal 100 displays all series information included in the designated examination ID as shown in an area 810 of FIG. 19 and prompts the user to select a series (step S560). In the area 800 of FIG. 19, the selected line is shaded. When the user selects a series to be interpreted, a medical image viewer is displayed by the display control unit 104 as shown in FIG. 20, and a series image with the specified series ID is displayed in a state where all slices can be paged (step). S570). The user performs image diagnosis while feeding slices. If the user is unsure of the image diagnosis, a similar case search is activated.

  The similar case search may be activated by using a predetermined shortcut key on the keyboard of the operation unit 102, or by right-clicking with the mouse to display the medical image viewer menu, and specifying the similar case search menu from the menu. Also good. When the start of similar case search is instructed, the management of the information terminal 100 is transferred to the ROI management unit 105 and the region of interest (ROI) is accepted. The user sets a region of interest (ROI) to the lesion on the slice of interest through the operation unit 102 (step S580). Here, a rectangle is set by dragging the mouse, and the input rectangle information (for example, the coordinates of the upper left point and the lower right point) is received by the ROI management unit and then passed through the communication control unit 110 to the case search system 300. To the communication control unit 304 (step S590). At the same time, among all the series images received in step S550 from the medical information management system 100, a slice image to be subjected to similar case search is also transmitted to the communication control unit 304 of the case search system 300 (step S600). Note that the slice ID may be used instead of the slice image itself to be searched for similar cases. In that case, the case search system 300 may acquire a slice image from the medical information management system 200 based on the slice ID.

  Next, a flow from when the case search system 300 searches for similar cases and initially displays the similar case search results on the information terminal 100 will be described.

  FIG. 21 is a sequence diagram showing a flow from the case search system 300 receiving a request for similar case search to returning the similar case search result to the information terminal until the initial screen display of the search result.

  The image feature extraction unit 302 of the case search system 300 extracts a predetermined multi-dimensional image feature (step S610).

  Examples of the “image feature amount” include those related to the shape of an organ or a lesion in a medical image, or those related to a luminance distribution. As the image feature amount, for example, Non-Patent Document 1 describes using a 490-dimensional feature amount. In the present embodiment, this image feature is used. Of course, other image features may be used.

  The similar case search unit 303 compares the image feature extracted by the image feature extraction unit 302 with the image feature (image feature data 4250 in FIG. 14) stored in the similar case data storage unit 301 for each region of interest ID. The distance is calculated (step S620). Then, similar cases whose distances are equal to or smaller than a predetermined threshold are sorted in ascending order of distance, and are determined as cases to be transmitted (step S630). Next, the communication control unit 304 includes a region of interest ID 4210, a slice ID 4220, an image finding ID 4230, a region of interest coordinate 4240, thumbnail image data 4260, and confirmation of the transmission target case among the similar case data stored in the similar case data storage unit. The disease ID 4400 for the number of diagnoses 4300 and the distance calculated by the similar case search unit 303 are transmitted for each region of interest ID by the number of regions of interest ID determined as the transmission target (step S640).

  As a similar case acquired as a result of the search, the distance is set to be equal to or smaller than a predetermined threshold. However, a predetermined number may always be acquired, or the number of cases always includes a predetermined number of disease names. For example, another determination method may be used.

  Hereinafter, generation of an initial screen for similar case search results (step S650) will be described with reference to the image finding system of FIG. 22 and the flowchart of FIG.

  FIG. 22 shows an image finding system used in the present invention. In the present invention, “image findings” are defined as words that express the image characteristics of lesions. The image finding system is expressed in a tree form from a superordinate concept to a subordinate concept as shown in FIG. Each image finding is composed of a notation and a corresponding image finding ID. For example, the notation that the image finding ID is IF0000 is “granular shadow”. The image finding system of FIG. 22 is stored in the image finding system storage unit 304 of FIG. Although the storage format will not be described in detail, any format may be used as long as the notation, ID, and upper layer ID are stored for each image finding.

  The highest hierarchy in the image finding system is “root” whose image finding ID is IF0.

  Note that FIG. 22 is an example of the image finding system, and other systems may be used as long as the image finding concept has a tree structure.

  The image finding ID 4230 in FIG. 14 described above stores one of the image finding IDs in the image finding system in FIG. When creating similar case data 4000, it is desirable for each case to have an image finding in the lowest hierarchy of the image findings system, but to give an image finding in the middle hierarchy in cases where it is difficult to distinguish It doesn't matter.

  FIG. 23 is a flowchart for generating an initial screen of similar case search results.

  First, in step S1000, the number of similar cases received is displayed in the search result number display area 712.

  Next, in step S1100, “all diseases” is displayed in the display condition display area 713 as an initial state.

  Next, in step S1200, the number of similar cases for each image finding ID in the image finding system of FIG. 22 is counted for all received similar cases. An example of the counted result is shown in FIG. In FIG. 24, the number of similar cases for the image findings is arranged as a number in a rectangle in the vicinity of the corresponding image findings. For example, it can be seen that the number of cases associated with “granular shadow” having an image finding ID of IF0000 is two. In this example, if one image finding is associated with one similar case, it can be seen that the number of similar cases received is 24.

  Next, in step S1300, the highest hierarchical image findings including all similar cases received in the lower hierarchy are specified. In the example of FIG. 24, the image finding ID is “lung parenchymal lesion” of IF00.

  Next, in step S1400, the image findings of a plurality of first division layers that are one layer below the image finding name of the highest layer identified in step S1300 are identified. Here, as a condition specified as an image finding of the first classification hierarchy, it is assumed that the received similar cases are included in itself and the lower hierarchy. In the example of FIG. 24, the “solid shadow” whose image finding ID is IF000 and the “ground glass shadow” whose image finding ID is IF001 are obtained.

  Next, in step S1500, thumbnail images, corresponding definitive diagnoses, and similarities are displayed in the similar case thumbnail display area 710 for each first classification hierarchy specified in step S1400. FIG. 25 is an example of a display image. The image findings of the first classification hierarchy identified in step S1400 are “solid shadow” and “ground glass shadow”, and FIG. 24 displays the cases in the lower hierarchy (total 17 cases and total 7 cases) of each image finding. To do. In the image finding system, when there are a case X corresponding to the image finding A and a case Y corresponding to the image finding B, and the image finding B is in the lower hierarchy of the image finding A, the case Y is assigned to the image finding A as “ It belongs to. In FIG. 25, since the number of displayable items in the horizontal direction is 5, the remaining cases are displayed by operating with the slide bars 714a and 714b.

  If the number of similar cases for each image finding ID is as shown in FIG. 26 in step S1200, the highest hierarchical image finding specified in step S1300 is “pulmonary parenchymal lesion” as in FIG. When the image finding of the first classification hierarchy is specified in S1400, “granular shadow” may be specified instead of “solid shadow”. The reason is that, in the case of FIG. 26, the only image finding where a similar case exists under the “solid shadow” is “granular shadow”. The screen in that case is as shown in FIG.

  Here, two types of display methods for cases belonging to the first classification hierarchy “solid shadow” and “ground glass shadow” will be described.

  The first is a method of selecting a case corresponding to an image feature having a predetermined variance or more in the image feature space. In FIG. 25, since the number of cases displayed in the horizontal direction is 5, five are selected from the display target case group. Specifically, the following method can be used. K-means clustering is performed in the image feature space for each image feature of the case group to be displayed, five cluster centers are determined, and a case closest to each cluster center is selected. An advantage of this display method is that variations of cases belonging to the same image finding classification can be grasped at a glance. In FIG. 24, the number of cases belonging to “solid shadow” is 14, but when there are more cases, only cases that look similar when displayed in order of similarity are displayed. I can't confirm. For a user who wants to grasp a variation of a case in an image finding at a glance, such an initial setting may be set, or the display method may be switched only in such a situation.

  The other is a method of displaying the search source image and the search result image in ascending order of the distance between the image feature amounts. In this case, since an image similar to the search source image is displayed, it is easy to determine the image findings and cases to be referred to.

  Finally, in step S1600, a disease name list is generated and displayed.

  First, a disease name list is generated from the received similar cases (here, NC cases). The disease name list management unit 106 generates a disease name list for NC similar cases based on the definite diagnosis (major classification disease name) 4700 and definite diagnosis (detailed classification disease name) 4800 attached to each of the NC similar cases. The generated disease name list is managed by the disease name list management unit 106 as table information as shown in FIG. In the disease name list, a disease name, a disease name ID (unique ID for each disease name), the number of cases, and a corresponding similar case ID (for the number of cases of the disease name) are managed. Then, the display control unit 104 causes the disease name list display area 730 to be displayed. Examples of display of the disease name list are shown in FIGS. 29, 30, and 31. FIG. FIG. 29 shows an example in which a definitive diagnosis is expressed by a detailed classification disease name, FIG. 30 shows an example in which a definitive diagnosis is expressed by a major classification disease name, and FIG. 31 shows an order in the number of cases for each major classification disease name. It is an example expressed for each number of detailed classification disease names included in each large classification disease name. In this case, the definitive diagnosis of the patient information storage unit 201 is expressed by a hierarchical structure of major classification disease names and detailed classification disease names. Steps S1000 to S1600 are executed by the display control unit 104.

  From the search result initial screen as shown in FIG. 25 or FIG. 27, the user narrows down the cases that serve as a reference for diagnosing the diagnosis target case 720 using the image finding display icon / selection button 711, and uses the disease name list 730. Thus, 1 to a small number of image diagnosis disease names of the diagnosis target case 720 are specified. Here, an example of FIG. 27 will be described.

  As mentioned in the problem to be solved by the invention, the present invention mainly provides a user who cannot select a reference case for diagnosing a case to be diagnosed from many thumbnail images as similar case search results. It is targeted. As shown in FIG. 27, the user displays similar cases as “granular shadows” and “ground glass shadows” in an easy-to-understand category in the image finding hierarchy, and a plurality of cases are displayed in each image finding. It is easy to determine which image finding corresponds to the diagnosis target case.

  When it is desired to find out more detailed image findings corresponding to the case to be diagnosed, a display icon / selection button 711 for image findings is selected.

  Here, an operation when the button 711a corresponding to “granular shadow” is pressed will be described. When the button corresponding to “granular shadow” is pressed, the image is divided into “non-lobular” and “lobular” which are the image findings of the lower layer of the granular shadow in FIG. 26 and displayed as shown in FIG. At this time, the number of cases display area 712 displays the number of cases whose image findings correspond to “non-lobular” and “lobular”. Furthermore, the disease name list 730 is also updated to a disease name list corresponding to cases whose image findings correspond to “non-lobular” and “lobular”.

  Since thumbnail images of only cases that fall under “non-lobular” or “lobular” are displayed from the similar cases that have been acquired, the user can determine which image findings correspond to the case to be diagnosed. It becomes easy.

  Further, when the similar case search performance is not sufficient, inappropriate cases are mixed in the search result cases, and as a result, the disease name list 730 also includes inappropriate disease names. In such a case, for a user who cannot select an appropriate case from the similar case search result, there is a risk of adversely affecting image diagnosis and learning while using this system. By narrowing down similar case search results with appropriate image findings using the present invention, the disease name list 730 also removes inappropriate disease names, so that such adverse effects can be reduced.

  In the process of narrowing down by image findings, as shown in FIG. 33, when the image finding display icon / selection button 711 is in the last hierarchy of similar cases (“lobular centrality” and “lobular marginality” in FIG. 26). Further, when the button 711 is further pressed, further detailed image findings cannot be developed, so that the disease name list 730 of similar cases corresponding to the pressed image findings is highlighted. As a highlighting method, enlarged display may be performed as shown in FIG. 34, the color may be changed, or the outer frame may be blinked. As shown in FIG. 35, in the case of a screen configuration in which the disease name list is not displayed from the beginning, the disease name list 730 may be displayed as shown in FIG.

  Note that a disease name list display button 750 may be provided as shown in FIG. 37 in case the user wants to refer to the disease name list even when the disease name list is not displayed from the beginning, even when the disease is not narrowed down to the final hierarchy.

  In the above description, the sum of the number of cases included in each image finding classification 711a, b,... Is displayed as the number of cases displayed in the similar case thumbnail display area 710 in the case number display area 712. The number of similar cases corresponding to the above may be displayed next to the button 711.

  In addition, although the example which extracts the image feature by the case search system 300 was shown above, you may implement on the terminal side. As described above, the user interface that presents the search results in the classification of image findings and classifies them in detail from the upper concept image findings to the lower concept image findings using a hierarchical image finding structure. Even if it exists, it becomes easy to judge the image finding of a search result, and it becomes easy to select the case used as a reference of a diagnosis as a result.

(Embodiment 2)
The second embodiment will be described below. Unless otherwise specified, the parts not described are the same as those in the first embodiment.

  In the present embodiment, it is allowed to assign a plurality of image finding IDs to one region of interest ID. For example, it is effective when a plurality of types of lesions are included in one region of interest as shown in FIG. The data structure of the similar case data 4000 in that case newly includes an image finding number 4225 as an entry as shown in FIG. 39, and image finding IDs corresponding to the number of image finding numbers 4225 are stored in the column of image finding ID 4230.

  Generation of an initial screen for similar case search results (step S650) in this embodiment will be described with reference to the flowchart of FIG.

  The operations in steps S1000, S1100, and S1600 are the same as those in the first embodiment. In step S1201, the number of similar cases for each “combination of image findings” is counted for all received similar cases. At this time, the result of counting the number of similar cases with respect to the image finding IDs for all received similar cases is obtained for each combination of image finding IDs. In the case of a combination of two image finding IDs, FIG. 24 is expanded to two dimensions, and in the case of a combination of three image finding IDs, it is expanded to three dimensions. The number of combinations of image findings to be handled may be determined in advance, and preferably 2-3. In the present embodiment, a combination of two image finding IDs will be described. A diagram obtained by multidimensional expansion of FIG. 24 is omitted.

  In step S 1301, the “image finding combination” in the highest hierarchy is specified. Considering the case where there are similar cases corresponding to all combinations of image findings except for root in the multidimensional expansion of FIG. 24, the “image finding combination” in the highest hierarchy is root (combination).

  In step S1401, the “combination of image findings” in the first classification hierarchy is specified. The result is a combination of all (ie, three) combinations of lung parenchymal lesions, bronchial lesions, and mediastinal lesions.

  In step S1501, a thumbnail image, a corresponding definitive diagnosis, and similarity are displayed in the similar case thumbnail display area 710 for each “combination of image findings” in the first classification hierarchy. Here, for simplification of illustration, only the display 711 of “combination of image findings” is shown in FIG. Three combinations identified in step S1401 are shown in 711a to 711c, respectively. Here, regarding the notation of the image findings that can be developed further in the lower hierarchy, an underline is added to indicate that the image can be expanded to the user. The “mediastinal lesion” is not underlined because there is no lower hierarchy in the image finding system of FIG. For the sake of simplification of explanation, it has been described that there is a similar case corresponding to a combination of all the image findings except for root. 42, there is a similar case corresponding to the combination of “bronchial lesion”, but there is no similar case corresponding to the combination of the image findings in the lower hierarchy of “bronchial lesion” and “pulmonary parenchymal lesion”. As such, “bronchial lesions” cannot be developed, and are therefore not underlined.

  Next, an operation for narrowing display cases using the display icon / selection button 711 for image findings will be described. Here, consider the example of FIG. In FIG. 41, “pulmonary parenchymal lesion” and “bronchial lesion” can be selected (ie, expanded). When “pulmonary parenchymal lesion” in 711a is selected, as shown in FIG. It is developed into a combination of three image findings that are subordinate to “lesion” and “bronchial lesion”. On the other hand, when “bronchial lesion” in 711a is selected, as shown in FIG. 44, the image is expanded into a combination of two image findings and “pulmonary parenchymal lesion” which are subordinate to “bronchial lesion”.

  As described above, by considering the combination of image findings, it is possible to handle a case where a plurality of types of lesions are included in one region of interest as shown in FIG. 38 while maintaining the effects of the first embodiment.

(Embodiment 3)
The third embodiment will be described below. Unless otherwise specified, the parts not described are the same as those in the first embodiment.

  In the present embodiment, similar case thumbnails displayed in the similar case thumbnail display area 710 are classified and displayed by image findings, and clinical information is also used when narrowing down by image findings. The “clinical information” here is information obtained by quantizing and structuring information of age 1300, gender 1400, medical history 1500, family history 1600, chief complaint 1700, and examination information 1800 in the patient information 1000 of FIG. Point to. When the content of the patient information 1000 is a free description, in order to use it for classification display or narrowing down in the similar case thumbnail display area 710, for example, for the age 1300, for example, “infant”, “infant”, “elementary school student”, A classification notation quantized and structured in advance, such as “high school student” and “30's”, and corresponding clinical information ID are defined, and these notations and IDs are used. For other items, for example, if the chief complaint 1700 or examination information 1800 is used, the words and descriptions having the same meaning but different notations are converted in advance into structured notations such as “abdominal pain” and “hotness”. Corresponding clinical information IDs are defined, and these notations and IDs are used.

  The data structure of the similar case data 4000 in that case newly includes a clinical information ID 4270 as an entry as shown in FIG. 45, and stores the clinical information ID.

  Next, the display when the display cases are narrowed down using the display icon / selection button 711 for image findings will be described. Here, consider the example of FIG. In FIG. 46, clinical information is displayed together with image findings. As with the first embodiment, the image findings can be expanded to lower-level image findings. By displaying and operating in this way, thumbnail images of similar case search results can be expressed by a combination of image findings and clinical information, and effective diagnosis support can be provided for diseases that cannot be diagnosed by image information alone. Yes.

  Note that disease progress information may be used instead of clinical information. As the degree of progression, for example, cancer stage information, severity classification of chronic kidney disease, and the like can be used. With such display and operation, thumbnail images of similar case search results can be expressed by a combination of image findings and disease progression, and effective diagnosis support can be performed.

  According to the present disclosure, the present invention can be used for a similar case search device that presents a similar case that serves as a reference in making a diagnosis using a medical image that is an interpretation target, an interpretation education device for a training interpretation doctor, and the like.

DESCRIPTION OF SYMBOLS 100 Information terminal 101 Display 102 Operation part 103 Input control part 104 Display control part 105 ROI management part 106 Disease name list management part 107 Communication control part 200 Medical information management system 201 Patient information storage part 202 Patient information management part 203 Medical image data storage part 204 Medical image data management unit 205 Diagnostic report management unit 206 Communication control unit 300 Case search system 301 Similar case data storage unit 302 Image feature extraction unit 303 Similar case search unit 304 Image finding system storage unit 400 Hospital network 501 Application 502 OS
503 memory

Claims (9)

A method for controlling an information terminal having a display connected to a case retrieval system for retrieving a medical image with reference to a medical image database,
(A) receiving a plurality of similar medical images having a predetermined degree of similarity with a medical image to be interpreted from the case search system;
Each of the plurality of similar medical images has additional information,
In the additional information of the similar medical image, a disease name and a first level image finding name are set,
The image finding name of the first hierarchy is associated with at least one image finding name of a hierarchy higher than the first hierarchy,
The image finding name of the layer higher than the first layer includes the image finding name of the highest layer,
The top-level image finding name is a single image finding name associated with all the first-level image finding names of the plurality of similar medical images,
(B) displaying a display screen including the plurality of similar medical images on the display;
The display screen is displayed by dividing the plurality of similar medical images by image finding names of a plurality of first division layers that are one layer below the image finding name of the highest layer,
(C) When selection of any of the image finding names of the plurality of first classification layers is detected,
(I) When the image finding name of the selected first division hierarchy is the first hierarchy set in the additional information of the similar medical image, the name of the selected first division hierarchy A display screen including all disease names set in the first layer image finding name associated with the image finding name is displayed on the display; (ii) the image finding of the selected first classification layer If the name one level lower than the first level is the image finding name of the second segment hierarchy that is higher than the first layer, the name is associated with the image finding name of the selected first segment hierarchy. A control method for displaying on the display a display screen in which a plurality of similar medical images are classified and displayed by image finding names of the second classification hierarchy. In the step (b), the image finding names of the plurality of first division layers are displayed so as to be selectable,
In the step (c), the selection of any of the image finding names of the plurality of first division layers is detected.
The control method according to claim 1. In the step (b), the display screen includes a display for accepting a user's disease name display request,
further,
(D) When the request for displaying the disease name of the user is detected, all the image finding names displayed on the display screen at the time of detecting the request for displaying the disease name of the user are associated with each other. Display a display screen including all disease names set in the image finding name on the display;
The control method according to claim 1. In the step (b),
A predetermined number of medical images selected from the plurality of medical images are displayed on the display screen,
A plurality of medical images having a predetermined variance or more in the image feature space are selected and displayed on the display screen from all the medical images corresponding to the image finding names of the plurality of first classification layers. ,
The control method according to claim 1. In the step (b), the plurality of medical images are in the order of similarity in the image feature space with the medical image to be interpreted, and are classified for each image finding name of the plurality of first classification layers. Displayed,
The control method according to claim 1. In the step (b), a similarity in which the first image finding name is associated with the plurality of similar medical images to a combination in which at least two of the plurality of first finding hierarchy image finding names are selected. When a predetermined number or more of medical images are included, combinations of image finding names of the plurality of first classification layers are also displayed separately.
The control method according to claim 1. The additional information of the similar medical image further includes at least one clinical information,
In the step (b), when the plurality of similar medical images include a predetermined number or more of similar medical images having additional information having a combination of the first layer image findings and the clinical information, The combination of one-layer image findings and the clinical information is also displayed separately.
The control method according to claim 1. In the additional information of the similar medical image, at least one disease progress information is set,
In the step (b), when the plurality of similar medical images include a predetermined number or more of similar medical images having additional information including a combination of the first-layer image findings and the disease progress information. The combination of the first layer image findings and the clinical information is also displayed separately.
The control method according to claim 1. further,
(D) In step (ii) of the step (c), after the display screen that is divided and displayed by the image finding name of the second divided hierarchy is displayed on the display, When the selection of an image finding name is detected,
(Iii) When the image finding name of the selected second classification hierarchy is the first hierarchy set in the additional information of the similar medical image, the name of the selected second classification hierarchy A display screen including all disease names set in the image finding name of the first hierarchy associated with the image finding name is displayed on the display; (ii) the image finding of the selected second classification hierarchy If the hierarchy one level below the name is the image finding name of the third classification hierarchy that is higher than the first hierarchy, it is associated with the image finding name of the selected second classification hierarchy Displaying on the display a display screen in which a plurality of similar medical images are classified and displayed by image finding names of the third classification hierarchy;
The control method according to claim 1.