Image diagnosis report preparation system

The integrated image diagnosis report system addresses inefficiencies in medical imaging by linking reading and reporting systems, allowing direct metadata input on the reading workstation, thereby reducing physician fatigue and enhancing reporting efficiency and data utilization.

JP2024162834APending Publication Date: 2024-11-21JAPANESE FOUND FOR CANCER RES
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
JP2023078759
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-11
Publication Date
2024-11-21

AI Technical Summary

Technical Problem

Existing medical imaging systems lack efficiency in recording and utilizing information from medical images, leading to physician fatigue, errors, and inefficiencies in creating diagnostic reports due to the separation of image viewing and reporting systems, which require frequent line-of-sight shifts and hinder secondary use of image data.

Method used

An integrated image diagnosis report creation system that links a reading workstation for medical images with a reporting system, allowing direct input of information as structured metadata on the reading workstation, reducing line-of-sight shifts and enabling efficient recording and secondary use of image data through AI-assisted diagnostics.

Benefits of technology

Enhances reporting efficiency by minimizing eye strain and errors, facilitating seamless information transfer, and enabling comprehensive data utilization and search capabilities through structured metadata input.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024162834000001_ABST
    Figure 2024162834000001_ABST
Patent Text Reader

Abstract

To provide a system that supports efficient and accurate preparation of an image diagnosis report and can facilitate a search for an image and similar cases.SOLUTION: An image diagnosis report preparation system comprises a reading work station that refers to a medical image and a reporting system, and allows a doctor to input items necessary for image diagnosis report preparation as meta data on the reading work station, to extract information necessary for image diagnosis without moving the eyes. When the doctor transmits the extracted information to the reporting system, the information is input by reporting software in accordance with the format of a predetermined image diagnosis report, which allows efficient preparation of an image diagnosis report.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a system for smoothly recording information obtained from an apparatus for observing or analyzing medical images in an image diagnosis report. [Background technology]

[0002] Medical images that visualize the inside of the human body are essential for the diagnosis and treatment of diseases. Not only CT and MRI, but also plain X-rays, angiograms, etc. are used as digital images. As the digitalization of medical images has become widespread, it has become easier to process imaging data by computer, and as a result, the speed of development of new diagnostic technologies has increased.

[0003] With the development of image diagnosis, the importance of image diagnosis reports prepared by imaging diagnosticians is rapidly increasing. However, a system for recording findings obtained by imaging diagnosticians from medical images has not yet been developed that can be said to be efficient enough. In addition, since a patient's past images are searched almost every time a report is prepared, efficiency and comprehensiveness are important, but it is difficult to say that the systems on the market meet the requirements in the field. Therefore, a system that dramatically improves the efficiency of report preparation is required.

[0004] In order to address the above-mentioned problems, methods for efficiently creating image diagnosis reports and methods for searching image data have been disclosed. Patent Document 1 discloses a system that can efficiently create reports by providing an inference engine for establishing building blocks that constitute data that specifies at least a part of the format and contents of a report. Patent Document 2 discloses a system that adds character information that constitutes a radiology report containing information about image data as metadata for searching image data, and can easily detect similar images and related information by keyword search.

[0005] However, although the method described in the cited document 1 may reduce the burden on doctors, it does not essentially improve the efficiency of doctors' work. In addition, the system described in the cited document 2 uses the information entered by doctors as metadata as is, so the doctor's choice of words and differences in description are reflected as is, which creates problems when performing precise searches. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Special Publication No. 2015-528959 [Patent Document 2] JP 2008-052544 A Summary of the Invention [Problem to be solved by the invention]

[0007] Since imaging devices such as CT and MRI and electronic medical record systems have been developed separately, the system for viewing the captured images and the system for recording the information read from the images are independent (see Figure 4(B)). The doctor observes the image on the software window of the reading system, moves his / her gaze to the software window of the reporting system displayed on a separate monitor to read the information from the image, and then enters the findings in the description field in this window (see Figure 4(A)). Therefore, when returning to image observation, it is necessary to switch the operation target from the reporting system to the reading system software and move his / her gaze. The task of observing medical images in this software window of the reading system and moving his / her gaze to the software window of the reporting system each time information is extracted requires the following steps each time information is extracted: interrupt observation → move the gaze → record the extracted information → return to the point where observation was interrupted and think → continue observation. The frequent gaze movement and interruption of thinking due to the inefficiency of this procedure leads to accumulation of fatigue in the doctor and is thought to increase the risk of errors, which is a safety problem. In addition, since the extracted information cannot be recorded on the reading system together with the image, it cannot be used for searches, which makes it difficult to lead to secondary use of the stored information.

[0008] In order to solve the above problems, the present invention aims to provide a system that can input information into medical records, image diagnosis reports, etc., in a device that refers to or analyzes medical images. In addition, the present invention aims to provide a system that can smoothly record and utilize the results of analysis performed by other systems, such as the analysis results verifying an artificial intelligence model or the output of the model. [Means for solving the problem]

[0009] The present invention is based on the information sharing between two subsystems that make up the image diagnosis report creation system: 1) a reading workstation that refers to medical images, and 2) a reporting system that creates image diagnosis reports. This information sharing realizes 1) a reduction in the burden of moving the line of sight, 2) a reduction in the burden of searching for past images, and 3) promotion of secondary use of image diagnosis report information.

[0010] The present invention relates to the following image diagnosis report creation system. (1) An image diagnostic report creation system comprising a leading workstation for referencing medical images and a reporting system for creating image diagnostic reports, wherein information extracted from the medical images by an authenticator is added to the medical images as metadata on the leading workstation. In this system, doctors can record information extracted from medical images on the reading workstation while viewing the medical images displayed on the reading workstation, which reduces the burden on doctors as they do not need to move their eyes to the software window of the reporting system each time they observe a medical image and extract information.

[0011] (2) An image diagnosis report creation system as described in (1), characterized in that the items entered on the reading workstation are transmitted to a reporting system and entered into specified items of the image diagnosis report by reporting software in the reporting system. Since the items entered on the reading station can be directly entered into the designated items of the image diagnosis report, the doctor only needs to check the contents according to the format of the image diagnosis report.

[0012] (3) An image diagnostic report creation system as described in (1) or (2), characterized in that the information extracted from the medical image by the certifier is input as metadata as structured information except for the free text entry areas. Because the information is structured, it is easy to search for the past history of the same patient or similar cases.

[0013] (4) An image diagnostic report creation system as described in (3), characterized in that it is equipped with a diagnostic support AI that detects anatomical structures from medical images, presents the names of organs and / or areas within the organs displayed in the medical images, detects abnormal conditions in the displayed medical images, and, if an abnormal condition is detected, presents candidate disease names. By incorporating diagnostic assistance AI, abnormal conditions will be detected by both the doctor and the AI, eliminating oversights and improving diagnostic accuracy. [Brief description of the drawings]

[0014] [Figure 1] FIG. 1 is a diagram showing an overview of a construction example of an embodiment of the present invention. [Diagram 2] 11 is a flowchart showing the flow of analysis processing of a diagnostic assistance AI. [Diagram 3] 1 is a flowchart showing the flow of creating an image diagnosis report by the reporting system. [Figure 4] FIG. 1 is a diagram showing a schematic diagram of conventional medical image referencing and medical record creation. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0015] In this specification, medical images refer to images obtained by medical procedures, such as plain X-rays, computed tomography (CT), magnetic resonance imaging (MRI), ultrasound tomography (US), nuclear medicine tests, angiography, endoscopy, and histopathology tests. In the following, a system that has been conventionally used to refer to these medical images is referred to as a reading system, and the medical image referencing system of this embodiment is referred to as a reading workstation to distinguish it from the reading system. In addition, in this specification, a report that records the findings of an image diagnosis is referred to as an image diagnosis report. An image diagnosis report is not only an independent report, but also includes medical records (electronic medical records) when information extracted from medical images is directly recorded in the medical records.

[0016] Standards for medical image data and systems for storing and managing the same can be those that are commonly used. Medical image data is stored or communicated according to the international standard Digital Imaging and Communications in Medicine (DICOM). As a result, information can be easily exchanged between products from multiple manufacturers. In addition, DICOM images contain associated patient information, modality, imaging device name, and the like as metadata. In the present invention, medical image data conforming to the DICOM standard is used unless otherwise specified.

[0017] A Picture Archiving and Communication System (PACS) can be used as a system for storing and accessing medical images. PACS is a system that receives medical image data taken with various modalities such as CT and MRI via a network, and stores and manages them, and has been introduced in many medical institutions both in Japan and overseas. This system can utilize existing standards and systems.

[0018] [Embodiment 1] First, the flow from viewing medical images using this system to creating an image diagnosis report will be explained (Fig. 1). Medical images output from an imaging device (10) are sent (110) to a PACS server (20), where they are stored, and when creating a report, they are sent (130) from the PACS server to a leading workstation (30) by the doctor's operation (120), and are displayed on a monitor in front of the doctor. The doctor detects whether the image is normal or abnormal on the monitor of the leading workstation. Using this system, the doctor can add information detected in the image to each image file as metadata. The information added includes the location information of the findings and the doctor's judgment (diagnosis, etc.) regarding the findings, as shown below.

[0019] The present embodiment is different from the conventional one in that the information can be input on the leading workstation. By changing the input destination from the conventional reporting system to the leading workstation, the burden on the doctor required to switch between operations and gaze directions to the reporting software can be eliminated.

[0020] Input on the Reading Workstation allows users to select from a number of options, making it possible to structure the data. In addition, content that does not fit into the options or templates can be written in a free-text field. In some cases, voice input can be used to automatically convert the voice data into text using AI and record it. Below, we will explain examples of items that can be entered on the Reading Workstation.

[0021] [Example of input items] (A) Authentication information regarding the person who inputs the output content (1) Whether or not it is certified, (2) Date and time of certification, (3) Certifier, (4) Certification history (revision history) The presence or absence of authentication is an item indicating whether the doctor has already performed authentication, i.e., image diagnosis, and is input together with the authentication date and time at the time the authenticator performs authentication. The identity of the authenticator (entering doctor), the input date and time (authentication date and time), and the time of revision of the input content (authentication history) are considered to have been completed by transferring the login information of the reporting system to the leading workstation system. In this case, the subject of authentication is the system. The time stamp is the time when the inputter performed the input completion operation. Alternatively, the authenticating doctor may select and input this information using tabs or the like. The authentication history is added, for example, when a follow-up is being performed and past images are referred to again and the content is changed. In this case, the person who made the change may be not only the same doctor, but also a different doctor.

[0022] (B) Abnormal negative information In this system, the area where the doctor judges the abnormality positive can be recorded on the reading workstation (see item C below), but the area and organs where the doctor judges the abnormality negative are also essential for medical treatment. In this system, the abnormality negative information is recorded in a structured manner.

[0023] (C) Location information of the abnormal area This system saves the position information of the abnormal area recorded by the doctor. The area that the doctor recognizes as abnormal can be surrounded and marked on the image with a line or the like. As a result, the outline and center coordinates of the abnormal area are recorded as metadata. In addition, the image itself is already assigned an image ID, such as an examination number, series number, image number, etc., as metadata. The doctor can select and input the organ name, organ area name, or section name that contains the abnormal area. In addition, after receiving an image from a medical device, if a diagnostic assistant AI is equipped that automatically extracts anatomical parts in the image, such as lungs, bones, liver, etc., and further detects abnormalities, the organ name that contains the abnormal area and the organ area name or section name that contains the abnormal area can be automatically presented as candidates, and the doctor only needs to confirm and input the organ name, organ area name, or section name that contains the abnormal area. By inputting the abnormal organ or area, the software narrows down the order of disease names that appear as disease candidates during diagnosis, which will be described later, and displays them in order of the number of cases in each area.

[0024] (D) Pixel value information of the pixels constituting the abnormal region In medical images, abnormal regions have unique pixel values. In this system, the pixel values ​​included in abnormal regions recorded by doctors are saved.

[0025] (E) Progression of abnormal areas (1) Linking information about abnormal regions in different images 1. List of anomalous areas, 2. Metadata of anomalous areas (2) Information on changes 1. Diameter, area, volume, 2. Pixel value (3) Time information of change In medicine, areas (lesions) that are judged to be "abnormal" continue to be observed regardless of whether medical intervention (treatment) is performed, and any subsequent changes and their details are recorded. In this system, if the image analysis calculation results indicate that an abnormal area recorded in a previous examination is the same as an abnormal area in the examination for which the report is being created, the two are linked. In addition, the system outputs the pixels contained in those abnormal areas and the changes over time in information derived from the pixel values. This pixel value-derived information refers to the statistical characteristics of the pixel values ​​and their arrangement. These statistical characteristics are used as representative values ​​of the pixel values ​​contained in the abnormal areas and are used to analyze changes over time.

[0026] (F) Record of physician's judgment regarding abnormal areas In this system, the doctor's judgment, including the diagnosis for abnormal areas, is recorded. Most of the judgment is recorded in a structured form. For those that do not fit into a structured form, free text input is allowed.

[0027] (G) Measures or proposed measures for abnormal areas (measures of examination, methods of follow-up, scheduled date and time of follow-up) In this system, in addition to the doctor's judgment of the abnormal area, the target or countermeasure is recorded. Most of it is recorded in a structured manner. If it is not suitable for structuring, free text input is also allowed.

[0028] After the above recording is completed on the reading workstation, the information added by the doctor is converted into metadata, and a new file format with the data added to the DICOM file, or a new DICOM file with the metadata written in the header of the DICOM file, is generated. The new file with the metadata added is stored on the PACS server (140). When the doctor wants to review the image at a later date, he or she can call up the image with the entered data added (150).

[0029] The information is entered into the image diagnosis report by sending it to the reporting system in response to a transmission command from the doctor (160). The transmitted information is automatically entered into each entry item of the image diagnosis report according to the format by the reporting software. The information entered on the reading workstation is structured information selected by selection tabs, etc., except for free-form writing fields. Meanwhile, the entry fields of the image diagnosis report and electronic medical record are also structured, so they are automatically converted into a predetermined format and entered. The doctor checks the documents created on the reporting system, fills in any further items that need to be filled in, and stores them in the structured medical database in the medical institution (170). Documents such as the image diagnosis report stored in the structured medical database can also be viewed by other medical professionals from their terminals (180, 190), making it possible to share patient information.

[0030] [Embodiment 2] This system displays a list of abnormal regions recorded in previous examinations. When an abnormal region is selected from this list, the image containing that abnormal region is retrieved and immediately displayed on the reading workstation. In addition, images containing the corresponding region are also displayed in the examination for which the report is being generated. This function allows doctors to comprehensively and quickly compare abnormal regions recorded in the past with changes up until the examination for which the report is being generated. Furthermore, statistical comparisons that take into account information derived from the pixels contained in the region are possible.

[0031] [Embodiment 3] In this system, it is possible to search for similar cases by searching for words written in the metadata. As mentioned above, the metadata is structured data because it is input by the doctor using options and templates. Therefore, when searching for similar cases, if you input keywords and search, it is possible to extract all the cases you are looking for. In addition, since the images are accompanied by information entered by the doctor when creating the image diagnosis report, such as the range of abnormal areas, the required information can be extracted immediately.

[0032] [Embodiment 4] The function of the diagnostic assistant AI is explained below (Figure 2). As mentioned above, by equipping this system with a diagnostic assistant AI, input at the reading workstation can be performed more efficiently. The diagnostic assistant AI may be installed inside the reading workstation or outside. When the doctor calls up an image to the reading station for authentication (S1), the diagnostic assistant AI, which has previously learned images of various regions and abnormal conditions that may occur in each region, is activated (S2). The diagnostic assistant AI detects which organ the target image belongs to and its anatomical structure (S3), determines the anatomical region (S4), and then automatically selects an appropriate diagnostic assistant AI to determine the abnormal state for each determined anatomical region and displays the output result of the AI. The results of the diagnostic assistant AI are displayed on the monitor of the reading station (S5), and the doctor checks the displayed results (S6), approves them if they are correct, and has them re-evaluated if they are not. If the displayed organ name, abnormal state, etc. are approved, the results are temporarily saved in an arbitrary location on the system. When the doctor subsequently inputs the items to be considered on the leading workstation, for example when inputting details of the abnormal area, options are displayed according to the determined anatomical area (S7). For example, if the diagnostic support AI determines that the area shown in the image is the "lungs", the options are programmed to display the names of diseases related to the "lungs" in order of the most common "lungs" disease names that have a high degree of match between the diagnostic support AI results and the features.

[0033] After the doctor has finished inputting all the information, the imaging diagnosis report can be created according to the metadata added by the doctor (Figure 3). The metadata added by the doctor is acquired by the reporting software (S11). The structured items of the imaging diagnosis report are automatically input by the reporting software from the metadata items input on the reading workstation (S12). However, the doctor must fill in the fields that should be filled in with free text. The diagnostic assistant AI is presented with the free text fields in advance, and is trained to extract what items to write and what to write. The diagnostic assistant AI extracts the free text fields (S13), extracts the items necessary for the report from the input fields (S14), and creates diagnostic findings (S15). The created findings are programmed to be added and revised by the doctor (S16) and then output to the designated area of ​​the imaging diagnosis report (S17). The doctor can finally check the imaging diagnosis report, make corrections if necessary, and complete the report. In addition, the diagnostic support AI learns from the parts of the free text that the doctor has corrected, so that it can output reports in the doctor's style. Therefore, ultimately, the doctor only needs to input the necessary information on the reading workstation, and then check the report created by the reporting software.

[0034] According to the above method, a doctor can refer to images and record findings simultaneously on a reading workstation, making image interpretation more efficient. Creating reports on the same screen not only saves time, but also prevents transcription errors when creating reports. In addition, it is possible to record various information as metadata together with images, making it possible to compare with past data and search for similar cases using the metadata.

Claims

1. An image diagnosis report creation system, comprising: The system is equipped with a reading workstation for referencing medical images and a reporting system for creating image diagnosis reports. An image diagnostic report creating system, characterized in that information extracted from a medical image by an authenticator is added to the medical image as metadata on a reading workstation.

2. Items entered on the Reading Workstation are sent to the reporting system, 2. The image diagnosis report creating system according to claim 1, wherein the information is input into a predetermined item of the image diagnosis report by a reporting software in the reporting system.

3. The information extracted from the medical image by the authenticator includes:

3. The image diagnosis report creation system according to claim 1, wherein information other than the free text entry area is input as metadata as structured information.

4. Detect anatomical structures in medical images; presenting the name of an organ and / or an area within the organ displayed in the medical image; The image diagnostic report creation system according to claim 3, further comprising a diagnostic assistant AI that detects abnormal conditions in the displayed medical image and, if an abnormal condition is detected, presents candidate disease names.

Citation Information

Patent Citations

  • Database system, program, image retrieval method, and report retrieval method

    JP2008052544A

  • A system and method for generating reports based on input from radiologists.

    JP2015528959A