The Value Of Mp-MRI Radiomics In Preoperative Prediction Of Extracapsule Extention Of Prostate Cancer

Objective: To investigate and validate the potential role of a radiomics signature in predicting the specific probability of extracapsular extension(ECE)of prostate cancer(PCa).Methods: From June 2013 to October 2019,the First Affiliated Hospital of Wannan Medical College carried out Mp-MRI examination and 101 cases of prostate cancer were confirmed by pathology(47 cases in the positive group and 54 cases in the negative group),using the radiant DICM viewer(64 bit)and ITK-SNAP(version 3.6)the software preliminarily processes each sequence of Mp-MRI and manually outlines ROI layer by layer,imports the sequence and voi information into AK software developed by GE company and extracts texture feature parameters,selects the best image group features by using the minimum redundancy maximum correlation(m RMR)and the minimum absolute contraction and selection algorithm(LASSO),and establishes the machine learning method of logistic Each sequence and joint sequence model,explore the relationship between each sequence and joint sequence model and pathological extracapsular extension state,and evaluate the differential diagnosis ability of each model through ROC analysis.Finally,we randomly assigned samples to the training group(n = 71)and the verification group(n = 30)for training and verification of each model.Results: 1.AUC value,sensitivity and specificity of T1 WI radiomics significance in training group and verification group were 0.96,0.911,0.892,0.78,0.692 and 0.705 respectively;AUC value,sensitivity and specificity of T2 WI radiomics significance in training group and verification group were 0.890,0.971,0.703;0.870,0.923 and 0.765respectively;AUC value and sensitivity of DWI radiomics significance in training group and verification group were 0.890,0.971 and 0.703 respectively The AUC value,sensitivity and specificity of DCE image in training group and verification group were 0.940,0.882,0.919;0.940,0.846 and 0.882,respectively.2.The AUC value,sensitivity and specificity of the combined sequence Radiomics model in the training and the verification group were 0.990,0.971,0.973;0.910,1.000 and 0.765,respectively.Through the analysis of decision curve,the clinical practicability of the model is confirmed.Conclusion: Based on the multiparametric magnetic resonance imaging(MpMRI),the Radiomics model has a good diagnostic efficacy for the prostate cancer extracapsular extension.It has great potential and clinical significance for predicting the prostate cancer extracapsular invasion before operation,so as to provide help for clinicians to develop personalized treatment plan.