Prediction Of Differentiation Degree Of Renal Clear Carcinoma Based On CT Imaging Omics Characteristics

Background: Renal cell carcinoma(RCC)is the most common subtype of renal cell carcinoma,while clear cell renal cell carcinoma is the most common type of renal cell carcinoma.In recent years,the preoperative diagnosis of renal clear cell carcinoma has received increasing attention.Traditional laboratory examination and imaging examination can investigate the lesion macroscopically,but can not find subtle lesion information.Although kidney biopsy can obtain the pathological information of lesion cells,it cannot dynamically observe the development of the lesion.Therefore,the combination of artificial intelligence and traditional imaging methods of imaging omics came into being.In the premise of non-invasive,better preoperative accurate prediction of renal clear cell carcinoma.Objective: In This this study,through the extraction and analysis of the imaging information of patients with renal clear cell carcinoma,a preoperative prediction model was established to better guide the clinical treatment and predict the prognosis.Methods: This study retrospectively selected between December 2018 and December 2020 of 102 patients with renal clear cell carcinoma as the research object,including all the objects of abdominal CT scan + 3 period enhance image,use MRMR after removing redundant and irrelevant features,use the lasso selection method finally out most the characteristics of forecast by using ROC curve and AUC image omics model is set up and verified,final diagnosis efficiency evaluation model.Results: A total of 10 most significant features were selected to establish an evaluation model.In this evaluation model,a total of 72 patients were trained.AUC was 0.85(95%CI: 0.75-0.92),sensitivity and specificity were 0.79 and 1.00,positive prediction rate and negative prediction rate were 1.00 and 0.66,respectively.The validation set consisted of 30 patients,of which the AUC was 0.90(95%CI:0.73-0.98),the sensitivity and specificity were 0.90 and 0.89,and the positive and negative predictive rates were 0.95 and 0.8,respectively.In this study,the predictive model can be used in a relatively safe area,and this model has good clinical application value.Conclusions: CT-based imaging omics model has a good effect on preoperative prediction of differentiation degree of renal clear cell carcinoma,which can play a certain guiding role in clinical work.