Prediction of Local Control in Brain Metastases following Stereotactic Radiosurgery Treatment

Radiomics shows promise for predicting local failure (LF) in brain metastases (BM) treated with stereotactic radiosurgery (SRS). This investigation sought to determine if radiomic features provide complementary information to routinely available clinical variables. We trained a Random Forest (RF) classifier for LF prediction on a dataset of 408 BMs from 87 patients in a cross-fold validation manner, to observe the change in predictive power from including radiomic features alongside clinical features. The addition of any one of the top nine radiomic features to the set of clinical variables resulted in a significant (pM.Sc.2021-11-21 00:00:0