Value of MR-T2WI and ADC maps texture analysis for the diagnosis of clinically significant prostate cancer

YUAN Zhuobin; LIU Bihua; ZOU Yujian; CHEN Zhaoting; FAN Xianmiao; CHEN Jin; WU Hongzhen

doi:10.20227/j.cnki.2096-3610.2025.05.006

Abstract：Objective To assess the diagnostic value of texture parameters based on T2-weighted imaging(T2WI)and apparent diffusion coefficient(ADC)maps for clinically significant prostate cancer(csPCa).Methods A total of 187 prostate cancer patients who underwent multiparameter magnetic resonance imaging(mpMRI)were enrolled.Participants were stratified into two groups:39 with Gleason score(GS)<7 and 148 with GS≥7.Texture features were extracted from T2WI and ADC maps using Omni-Kinetic post-processing software and compared between groups.Receiver operating characteristic(ROC)curve analysis assessed the diagnostic performance of texture parameters.Binary logistic regression identified independent predictors of GS≥7.Results Ten texture parameters were calculated for T2WI and ADC maps,including first-order statistics(mean,variance,skewness,energy,entropy,uniformity,kurtosis)and second-order statistics(GLCM energy,GLCM entropy,and GLCM correlation).The differences in T2WI texture parameters mean,variance,kurtosis,GLCM energy,GLCM entropy,and GLCM correlation between prostate cancer with GS<7 group and GS≥7 group were statistically significant.The AUCs of GLCM correlation,GLCM energy,and GLCM entropy are relatively larger,indicating high sensitivity.There was no statistically significant difference between the two groups of skewness,energy,entropy,and uniformity.The texture parameters mean,skewness,uniformity,kurtosis,GLCM energy,GLCM entropy,and GLCM correlation of ADC images showed statistically significant differences.The AUC of mean and uniformity is relatively large,with uniformity having the highest sensitivity and mean having the highest specificity;there was no statistically significant difference in the texture parameters of variance,energy,and entropy between the two groups.Binary logistic regression analysis showed that T2WI texture parameter GLCM entropy and ADC texture parameters uniformity,GLCM energy,and GLCM entropy can used as influencing factors for csPCa identification.When T2WI texture parameters GLCM entropy and ADC texture parameters uniformity,GLCM energy,and GLCM entropy were combined,the diagnostic efficiency is optimal.Conclusion Texture parameters from T2WI and ADC maps serve as non-invasive biomarker for diagnosing csPCa.