Application Of MRI-based Radiomics In The Differential Of Parotid Pleomorphic Adenoma And Warthin Tumor

Objective: Preoperative differentiation between pleomorphic adenoma and Warthin tumor of parotid gland is important for treatment decisions.In this study,we aim to develop and validate an MRI-based model to differentiate between pleomorphic adenoma and Warthin tumor of parotid gland.Methods: This retrospective study included 242 patients with pleomorphic adenoma(n=131)or Warthin tumor(n=111)that were randomly divided into training cohort and validation cohort(7:3 ratio).Radiomics features were extracted from T1-weighted and fat-suppressed T2-weighted images,and clinical characteristics were gathered to build a clinical model.After extracting the radiomics features on MRI images,we constructed radiomics signature models using a multivariable logistic regression,random forest and support vector machine in the training cohort and evaluated its performance in the validation cohort.A combined model was also built using both the best radiomics signature and clinical characteristics.The performance of all the models was evaluated and validated using ROC curves on the training and validation cohorts.Results: The best radiomics signature was developed based on support vector machine combined MRI images.The classification accuracy of the combined model well differentiated between pleomorphic adenoma and Warthin tumor for both the training and validation cohorts(93.45% and 90.54%,respectively).The area under the receiver operating characteristic curve(AUC)of the combined model was significantly better than that of the clinical model for both the training(0.984 vs.0.845,p < 0.001)and validation cohort(0.974 vs.0.853,p = 0.002)cohorts.Conclusion: Our MRI-based combined model performed well for differentiating pleomorphic adenoma of parotid gland from Warthin tumor,and may help the clinical decision-making process.