Research On Structured Method Of Pathology Report Based On Deep Learning

With the development of medical information technology in China,hospitals have accumulated a large amount of unstructured data when providing medical services to patients.As one of the important unstructured documents,the pathology report is mainly text format data written by pathologists using natural language,recording the basic information of patients,gross inpection,microscope,pathology diagnosis.The text in the pathology report is essential for the physician to make a diagnosis,and is an important basis for clinical diagnosis and treatment,as well as being of great value for pharmaceutical and medical research.However,these pathology data are underutilized because of unstructured nature.Therefore,in this paper,a deep learning model,the bidirectional long and short term memory network(Bi LSTM)and conditional random field(CRF),is used to extract the text feature labels for structured design a structured processing system to support the specimens and extraction of indicator values.This paper first analyzes the textual data structure of pathology reports and summarizes the structure of the report's textual content.Then this paper constructs a professional pathology dictionary and a structured process for pathology reports.Through the testing of the real data set,the accuracy of the algorithm P is 94.5%,the average recall rate R is 96.7%,and the average F1 value is 0.96.Finally,this paper introduces the algorithm system implementation of the text structure of the pathology report,and the specific application in the pathology information project.The system can effectively improve the efficiency of pathologyists' diagnosis,assist pathology staff to be able to file or search pathology slides and whole image slides efficiently,and can provide some important data to clinical reaserch which inludes pathological analysis and drug development.Big data in pathology medicine will become the trend,and instead of looking at the slides under a microscope,doctors may view whole slide image.The diagnostic aid system will suggest the pathology diagnosis,and the doctor will have structured statements to choose from when writing the report.Once the diagnosis is complete,the digital slides are automatically coded and categorized for easy retrieval,and the structured data from the report is stored and used for analysis.Pathology text structuring is constantly being optimized based on samples,classifications,and needs.