Study On The Application Of Imaging Histology Based On CT Images In The Diagnosis Of Pulmonary Embolism

Objective To explore the feasibility and value analysis of the application of imaging histological features in lung embolism diagnosis based on C T images,and to provide new ideas for clinical diagnosis of pulmonary embolism.Methods Retrospective analysis of CTPA images of 56 patients wi th pulmonary embolism that met the criteria for inclusion from February 2018 to May 2019.These included a total of 34 cases of pulmonary embolism above the stage level and 22 cases of pulmonary embolism at the stage level.Step:(1)observe the presence of embolisms in the pulmonary arteries on the CTPA image,and record the number,location and type of embolisms,(2)all incoming case images are imported into ITK-SNAP software,select the maximum level of the embolized arterial blood supply area,manually sketch the area of interest(Region of Interest,ROI)as the case group,and then select the normal position c orresponding to the opposite side,and draw the same area of interest(ROI)with the same area of lesions as the control group.(3)Using PyRadio mics analysis software to extract image group features,and after the characteristic sweding,three more discer ning group features were obtained.40 cases were randomly selected as training sets and 16 cases as validation sets.(4)Using multi-factor logic regression analysis method,the predictive model of pulmonary embolism is established in the training set and verified in the validation set.Build the ROC curve and calculate the prediction ability of the under-curve area(AUC)evaluation model.Results According to the case group and control group,a total of 660 imaging group features were extracted,and the t hree imaging group features were finally obtained by feature screening,which were the most predictive value,respectively,the original variatio n of the peak,bias,and grayscale symbiotic matrix(GLCM)based on the hetogram features,based on the above T he AUC values of the models established by the three features are 0.779(95% CI,0.68-0.879),0.742(95% CI,0.57-0.915)in the training set and validation set.Conclusion(1)As a new noninvasive prediction method,the imaging group model based on CT image shows good classification accuracy for the pulmonary embolism group and the control group.(2)The model based on the original variation of h istogram features in the image group is well predicted.(3)Based on the results of this study,the imaging his tological analysis based on CT images is expected to provide quantitative parameters for the diagnosis of peripheral pulmonary embolism,thus gui ding the clinical development of personalized treatment programs.