Radiomics-based machine learning model to predict risk of death within 5-years in clear cell renal cell carcinoma patients

The aim of this study was to develop radiomics–based machine learning models based on extracted radiomic features and clinical information predict the risk death within 5 years for prognosis clear cell renal carcinoma (ccRCC) patients. According image quality data availability, we eventually selected 70 ccRCC patients that underwent CT scans. Manual volume-of-interest (VOI) segmentation each performed by an experienced radiologist using 3D slicer software package. Prior feature extraction, pre-processing images extract different features, including wavelet, Laplacian Gaussian, resampling intensity values 32, 64 128 bin levels. Overall, 2544 radiomics were from VOI patient. Minimum Redundancy Maximum Relevance (MRMR) algorithm used as selector. Four classification algorithms used, Generalized Linear Model (GLM), Support Vector Machine (SVM), K-nearest Neighbor (KNN) XGBoost. We Bootstrap method create validation sets. Area under receiver operating characteristic (ROC) curve (AUROC), accuracy, sensitivity, specificity assess performance models. best single among 8 achieved XGBoost model a combination (AUROC, with 95% confidence interval 0.95–0.98, 0.93–0.98, 0.93–0.96 ~1.0, respectively). developed robust radiomics-based classifier is capable accurately predicting overall survival RCC This signature may help identifying high-risk who require additional treatment follow up regimens.