Study Of MRI Texture Analysis To Differentiate Diagnosis Between Borderline And Malignant Epithelial Ovarian Tumors

Objective: To explore the value of magnetic resonance imaging(MRI)-based whole tumor texture analysis in differentiating borderline epithelial ovarian tumors(BEOTs)and FIGO stage Ⅰ / Ⅱ malignant epithelial ovarian tumors(MEOTs).Materials and Methods: A total of 88 patients with histopathologically confirmed ovarian epithelial tumors after surgical resection,including 30 BEOT and 58 FIGO stageⅠ / Ⅱ MEOT patients were divided into a training set(n = 62)and a test set(n = 26).The clinical and conventional MRI features were retrospectively reviewed.The texture features of tumors based on T2-weighted imaging(T2WI),diffusion-weighted imaging(DWI)and contrast-enhanced T1-weighted image(CE-T1WI)were extracted via Ma Zda software,and three top weighted texture features were selected by Random Forest algorithm.A non-texture logistic regression model in the training set was built to include those clinical and conventional MRI variables with P < 0.10.Then,a combined model integrating non-texture information and texture features was built in the training set.The model assessed in the training set was applied to the test set.Finally,Receiver Operating Characteristic(ROC)curves were used to assess the diagnostic performance of the models.Results: The combined model showed better performance in categorizing BEOTs and MEOTs(sensitivity,92.5%;specificity,86.4%;accuracy,90.3%;AUC,0.962)than nontexture model(sensitivity,78.3%;specificity,84.6%;accuracy,82.3%;AUC,0.818),the AUCs was statistically difference(P = 0.038).In the test set,the AUCs,sensitivity,specificity and accuracy were 0.840,73.3%,90.1% and 80.8% of the non-texture model and 0.896,75.0%,94.0% and 88.5% of the combined model.There was no significant difference in AUC between the training set and the test set of the combined model(P =0.348).The comparison and analysis between logistic regression model and single texture feature showed that the diagnostic efficiency of texture feature with the best discrimination ability extracted from T2 WI is equivalent to that of non-texture model,and the diagnostic efficiency of combined model was significantly better than that of any single texture feature.Conclusion: MRI-based texture features combined with clinical and conventional MRI features could assist in characterizing the differences between BEOT and FIGO stage Ⅰ /Ⅱ MEOT patients.