Design And Implementation Of Pneumonia Assistant Diagnosis System Based On Multimodality

Pneumonia is usually an infectious disease of the lungs caused by viruses,bacteria and other similar pathogens,which poses a greater threat to the human body.In December 2019,a new type of coronavirus pneumonia(Coronavirus disease 2019,COVID-2019)broke out and spread across the world.The rapid and accurate diagnosis of pneumonia has become an important research field at present.The diagnosis methods of pneumonia generally include viral testing,blood routine,imaging screening,etc.,but there are problems such as redundant methods,long diagnosis time,and high missed detection rate,and the work of reading doctors is heavy.Therefore,the use of new artificial intelligence technology to assist doctors in the preliminary screening of pneumonia patients has important theoretical research significance and practical application value.The main research content of the thesis includes the following parts:(1)The current status of pneumonia diagnosis technology at home and abroad and related algorithms such as lesion screening and pneumonia diagnosis based on radiographic images are analyzed.(2)Aiming at the problems of high missed detection rate and long diagnosis time in the single radiographic examination method,a multi-modal pneumonia auxiliary diagnosis plan based on text examination results and computer tomography(CT)images was designed.The text examination results include Etiology test report,patient’s past medical history.(3)Researched the GIVMT(Google Inception V3 MLP Text)model of pneumonia diagnosis based on CT images as the core and fusion of pathogenic test reports and patient’s past medical history information.Based on the above ideas,a multi-modal pneumonia auxiliary diagnosis system is designed and implemented using Python and Bootstap.The auxiliary diagnosis system displays and processes CT image data,supports DICOM3.0,and realizes data preprocessing,multi-planar reconstruction(MPR),Lesion screening,follow-up comparison and other functions.The system automatically recognizes and calculates the DICOM image data transmitted by CT equipment or medical image storage and transmission system(PACS),combines the test report and past medical history of the same patient,optimizes the calculation result,and provides it as a visualized diagnosis and treatment suggestion for diagnosis and treatment Doctors,assisting doctors to complete the diagnosis and treatment of the film.Using the average method to calculate the reading conditions of 10 doctors in a hospital before and after using the system,the average daily reading volume increased from45.2 to 60.3,which improved the efficiency of reading.