CN-115758993-B - Physical examination report analysis method, device and system and readable storage medium

CN115758993BCN 115758993 BCN115758993 BCN 115758993BCN-115758993-B

Abstract

A physical examination report analysis method, a physical examination report analysis device, a physical examination report analysis system and a readable storage medium are applied to the technical field of information detection. The method comprises the steps of carrying out page splitting processing on a physical examination report to obtain a plurality of physical examination report pages, carrying out word recognition on each physical examination report page, obtaining word recognition results, extracting medical text information, and determining medical text groups corresponding to the medical text information according to preset connection relations. The application reduces the difficulty of medical long-term analysis, can be directly applied to a review system of a nuclear maintenance worker and a client health portrait system, can obtain information structured output with high accuracy and high robustness aiming at physical examination reports of different formats of different institutions, and improves the nuclear maintenance work efficiency.

Inventors

QIN YAFEN
GAO CHAO
MAO GUOQING

Assignees

太保科技有限公司

Dates

Publication Date: 20260505
Application Date: 20221104

Claims (7)

1. A method for analyzing a physical examination report, comprising: Carrying out page splitting processing on the physical examination report to obtain a multi-page physical examination report page; performing word recognition on each physical examination report page to obtain word recognition results of each physical examination report page; Extracting medical text information according to the text recognition result of the physical examination report page of each page, wherein the medical text information comprises an examination item name, an examination item result, an examination item unit and a normal range value; Determining a medical subject category from the medical text information, the medical subject category including surgical examination, ophthalmic examination, pediatric examination, ultrasound examination, and gynecological examination; according to the position relation of each medical text message, enhancing the spatial position representation among the medical text messages in a relative position coding mode, and learning the connection relation of each medical text message by using a deep learning method to determine a medical text group corresponding to each medical text message, wherein the medical text group comprises a plurality of parallel medical text groups under the same medical theme type; according to the connection relation of the medical text information, determining medical text groups corresponding to the medical text information, wherein each medical text group consists of the medical text information with the connection relation, and the medical text groups comprise a plurality of parallel medical text groups under the same medical theme class; And obtaining the structured output of the physical examination report by determining the medical topic category corresponding to each medical text group, wherein the structured output comprises the medical topic category, the medical text group and the medical text information.
2. The method of claim 1, wherein the paging out the physical examination report comprises: Carrying out page splitting processing on a physical examination report with a storage format of TIFF through file analysis; and (3) carrying out page splitting processing on the physical examination report with the storage format of PDF through an automatic tool or an open source tool.
3. The method of claim 1, wherein said text identifying each of said physical examination report pages comprises: And performing character recognition on each physical examination report page by utilizing an OCR method.
4. The physical examination report analyzing device is characterized by comprising a page disassembly module, a character recognition module, an extraction module and a determination module; the page splitting module is used for carrying out page splitting processing on the physical examination report to obtain a multi-page physical examination report page; The character recognition module is used for carrying out character recognition on each physical examination report page and obtaining a character recognition result of each physical examination report page; the extraction module is used for extracting medical text information according to the text recognition result of the physical examination report page of each page, wherein the medical text information comprises an examination item name, an examination item result, an examination item unit and a normal range value; The medical text information detection system comprises a determination module, a detection module and a structural output module, wherein the determination module is used for determining medical topic categories according to medical text information, the medical topic categories comprise surgical examination, ophthalmic examination, pediatric examination, ultrasonic examination and gynecological examination, the spatial position representation among the medical text information is enhanced in a relative position coding mode according to the position relation of each medical text information, the connection relation of each medical text information is learned by a deep learning method to determine medical text groups corresponding to each medical text information, the medical topic categories comprise a plurality of parallel medical text groups, the medical text groups corresponding to each medical text information are determined according to the connection relation of each medical text information, each medical text group comprises a plurality of parallel medical text groups under the same medical topic category, and the structural output of the physical examination report comprises the medical topic categories, the medical text groups and the medical text information.
5. The device according to claim 4, wherein the page splitting module is specifically configured to: Carrying out page splitting processing on a physical examination report with a storage format of TIFF through file analysis; and (3) carrying out page splitting processing on the physical examination report with the storage format of PDF through an automatic tool or an open source tool.
6. The physical examination report analyzing device is characterized by comprising a memory and a processor; the memory is used for storing programs; the processor being configured to execute the program to implement the steps of the method according to any one of claims 1 to 3.
7. A computer storage medium on which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 3.

Description

Physical examination report analysis method, device and system and readable storage medium Technical Field The present application relates to the field of information detection technologies, and in particular, to a method, an apparatus, a system, and a readable storage medium for analyzing a physical examination report. Background The physical examination report refers to a document with a certain format generated from data of physical reaction by examining a body. In the existing physical examination process, various physical examination data are required to be summarized to generate a report. In order to facilitate the insurance company to know the physical condition of the insurance company in detail before the insurance applicant makes an insurance application, whether to accept the insurance application or select a proper dangerous seed for the insurance applicant design is finally determined, and the information of the insurance applicant's physical examination list needs to be extracted and applied. The information reported by early physical examination needs to be manually extracted and analyzed, the efficiency is quite low, and the long-time recognition and extraction process can cause eye fatigue. Along with the progress of science and technology, the fields of medical treatment and insurance are gradually digitalized, and information of physical examination reports can be automatically extracted based on intelligent vision technology, but in the prior art, only one-to-one relationship in groups can be represented, and one-to-many relationship cannot be represented. For example, in the physical examination report, there are two examination items of left eye vision and right eye vision, and in the prior art, only one examination item can be represented, but a one-to-many relationship cannot be formed. Disclosure of Invention The application provides a physical examination report analysis method, a physical examination report analysis device, a physical examination report analysis system and a physical examination report analysis readable storage medium, which can obtain high-accuracy and high-robustness information structured output aiming at physical examination reports of different formats of different mechanisms. The application discloses the following technical scheme: In a first aspect, the present application provides a method for analyzing a physical examination report, where the method includes: Carrying out page splitting processing on the physical examination report to obtain a multi-page physical examination report page; performing word recognition on each physical examination report page to obtain word recognition results of each physical examination report page; Extracting medical text information according to the text recognition result of each physical examination report page; And determining medical text groups corresponding to the medical text information according to a preset connection relation. In some possible implementations, the splitting the page of the physical examination report includes: Carrying out page splitting processing on a physical examination report with a storage format of TIFF through file analysis; and (3) carrying out page splitting processing on the physical examination report with the storage format of PDF through an automatic tool or an open source tool. In some possible implementations, the text recognition on each page of the physical examination report page includes: And performing character recognition on each physical examination report page by utilizing an OCR method. In some possible implementations, the method for learning the connection relationship includes: and according to the position relation of the medical text information, deeply learning the connection relation of each medical text information, wherein the position relation is represented by a relative position coding mode. In some possible implementations, the medical text information includes an examination item name, an examination item result, an examination item unit, and a normal range value. In a second aspect, the application provides a physical examination report analyzing device, which comprises a page disassembly module, a character recognition module, an extraction module and a determination module; the page splitting module is used for carrying out page splitting processing on the physical examination report to obtain a multi-page physical examination report page; The character recognition module is used for carrying out character recognition on each physical examination report page and obtaining a character recognition result of each physical examination report page; The extraction module is used for extracting medical text information according to the text recognition result of each physical examination report page; the determining module is used for determining medical text groups corresponding to the medical text information according to a preset connection relation. In so