Interpretable Multiparametric MRI Radiomics-based Machine Learning Model for Preoperative Differentiation between Benign and Malignant Prostate Masses: A Diagnostic, Multicenter Study

Wenjun Zhou¹Zhangcheng Liu¹Jindong Zhang¹Shuai Su¹Yu Luo¹Lincen Jiang¹Kun Han¹Guohua Huang²Jue Wang³Jianhua Lan²Delin Wang^4*

• ¹Department of Urology, First Affiliated Hospital of Chongqing Medical University, Chongqing, China
• ²Guang'an People's Hospital, Guang'an, China
• ³Panzhihua Central Hospital, Panzhihua, China
• ⁴First People's Hospital of Chongqing, Chongqing, China

Objective: To develop and externally validate multiparametric MRI (mpMRI) radiomics-based interpretable machine learning (ML) model for preoperative differentiating between benign and malignant prostate masses.Patients who underwent mpMRI with suspected malignant prostate masses were retrospectively recruited from two independent hospitals between May 2016 and May 2023. The prostate mass regions in T2WI and DWI MRI images were segmented by the ITK-SNAP. The 'PyRadiomics' was utilized to extracted radiomics features. The inter-and intra-observer correlation analysis, T test, Spearman correlation analysis, and the least absolute shrinkage and selection operator (LASSO) algorithm with a five-fold cross-validation were applied for feature selection. Five ML learning models were built using the chosen features. Model performance was evaluated with internal and external validation, using area under the curve (AUC), Calibration curves, and Decision curve analysis to select the optimal model. The interpretability of the most robust model was conducted via the SHAP.: A total of 567 patients were enrolled, consisting of the training (n=352), internal test (n=152), and external test (n=63) sets. 2632 radiomics features were extracted from ROIs of T2WI and DWI images, which were reduced to 18 via the LASSO. Five ML models were established, among which the RF model presented the best predictive ability, with AUCs of 0.929 (95% confidential interval [CI]: 0.885-0.963) and 0.852 (95 % CI: 0.758-0.934) in the internal and external test sets, respectively. The Calibration and Decision curve analysis confirmed the excellent clinical usefulness of the RF model. Besides, the contributing relations of the radiomics features were uncovered using the SHAP. Conclusions: Radiomics features from mpMRI combined with machine learning facilitates accurate preoperative evaluation of the malignancy in prostate masses. The SHAP can disclosed the underlying prediction process of the ML model, which may promote its' clinical applications.