Multiparametric MRI-based radiomics for predicting the EGFR mutation status in patients with non-small cell lung cancer

Motivation: Predicting epidermal growth factor receptor (EGFR) mutation status in non-small cell lung cancer(NSCLC) facilitates the selection of clinical treatment options. Goal(s): To predict EGFR mutation status in NSCLC using MRI-based radiomics and explore the optimal sequence for prediction. Approach: 117 patients (72 with mutant, 45 with wild-type) were included. ZTE, T2W-FS, and DWI sequences were selected for radiomics features extraction. The abilities of the features to predict EGFR mutation status were analyzed and compared. Results: The radiomics signature of DWI showed the best diagnostic performance in both the training and validation cohorts , which yielded an AUC of 0.873 in the validation cohort. Impact: DWI radiomics signature can be used as a noninvasive tool for predicting EGFR mutation status in NSCLC, which is helpful to guide therapeutic strategies.