Targeted T2-FLAIR Dropout Training Improves Robustness of nnU-Net Glioblastoma Segmentation to Missing T2-FLAIR

Purpose: To determine whether targeted T2 fluid-attenuated inversion recovery (T2-FLAIR) dropout training improves glioblastoma MRI tumor segmentation robustness to missing T2-FLAIR without degrading performance when T2-FLAIR is available. Materials and Methods: This retrospective multi-dataset study developed nnU-Net models on BraTS 2021 (n=848) and externally tested them on UPenn-GBM glioblastoma MRI (n=403; 2006-2018; age 18-89 years; 60% male). Models were trained with no dropout or targeted T2-FLAIR dropout (probability rate r=0.35 or 0.50) by replacing only the T2-FLAIR channel with zeros. Inference used T2-FLAIR-present and T2-FLAIR-absent scenarios (T2-FLAIR set to zero). The primary endpoint was Dice similarity coefficient (DSC); secondary endpoints were 95th percentile Hausdorff distance and Bland-Altman whole-tumor volume bias. Equivalence was assessed with two one-sided tests using +/-1.5 DSC percentage points, and noninferiority versus HD-GLIO used a -1.5-point margin. Results: With T2-FLAIR present, median overall DSC was 94.8% (interquartile range, 90.0%-97.1%) with dropout and 95.0% (interquartile range, 90.3%-97.1%) without dropout (equivalence supported, p<0.001). With T2-FLAIR absent, median overall DSC improved from 81.0% (interquartile range, 75.1%-86.4%) without dropout to 93.4% (interquartile range, 89.1%-96.2%) with dropout (r=0.35); edema DSC improved from 14.0% to 87.0%, edema 95th percentile Hausdorff distance improved from 22.44 mm to 2.45 mm, and whole-tumor volume bias improved from -45.6 mL to 0.83 mL. Dropout was noninferior to HD-GLIO under T2-FLAIR-present (all p<0.001). Conclusion: Targeted T2-FLAIR dropout preserved segmentation performance when T2-FLAIR was available and reduced segmentation error and whole-tumor volume bias when T2-FLAIR was absent.