Research On Structured Processing Of Medical CT Report Texts

The medical CT report text,as an important part of the electronic medical records,is a necessary reference for doctors to diagnose their illnesses and treat them symptomatically.The CT report consists of a description of the image and a diagnosis based on the findings.The description of the images is the unstructured data described by the imaging physician using a standardized narrative text language based on CT scan images.Diagnostic recommendations are the subjective descriptions of the doctor's experience in the textual description of the images,combined with their own reading experience.The essence of this judgment process is to manually extract the core content of the CT image and process the text structure.Based on the characteristics of descriptive text data seen in the image of the CT report,the paper uses the techniques of Chinese word segmentation and cluster analysis to study the structure of the text data of the CT report,and realizes the structural extraction and analysis of the CT report text,in order to enhance the CT report.Text data utilization provides technical support.Based on the data characteristics of the medical CT report text,the thesis gives the main research content of the paper,expatiates on the difficulties and solutions of the structure research of CT report text,and designs the overall framework of the study.The overall framework of the specific processing module includes:CT report text preprocessing module,build custom medical dictionary module and data structure conversion module.The CT report text preprocessing module mainly performs clause segmentation and special word marking on the original text.Taking the result of the preprocessing module as the object of processing,building a custom medical dictionary module to extract key information using text cluster analysis and other methods,and expanding the original medical thesaurus.The data structure conversion module uses the custom medical dictionary as the basis for word segmentation processing,and uses the structured conversion algorithm to extract keyword pairs in the form of text data key value pairs,and combines negative word detection and redundant vocabulary filtering to obtain a satisfactory CT report.Structured text results.The paper randomly selected the medical CT report text data,set up experimental research data sets,collected theoretical analysis and experimental optimization,and obtained the best parameter threshold.After selecting the optimal experimental parameters,the custom medical dictionary and the structured algorithm are compared with the general word segmentation software and the dependency syntax structure algorithm by the manual calculation method,which verifies that the method used in this paper can achieve the desired effect.The research results of the dissertation overcome the unavailability of generic word segmentation software in the medical field,realize the structured extraction of medical CT report text data,and provide reliable data support for medical data analysis.