Integrating Deep Learning and Radiogenomics: A Novel Approach to Glioblastoma Segmentation and MGMT Methylation Prediction.

Radiogenomics, which integrates imaging phenotypes with genomic profiles, enhances diagnosis, prognosis, and treatment planning for glioblastomas. This study specifically establishes a correlation between radiomic features and MGMT promoter methylation status, advancing towards a non-invasive, integrated diagnostic paradigm. Conventional genetic analysis requires invasive biopsies, which cause delays in obtaining results and necessitate further surgeries. Our methodology is twofold: First, an enhanced U-Net model segments brain tumor regions with high precision (Dice coefficient: 0.889). Second, a hybrid classifier, leveraging the complementary features of EfficientNetB0 and ResNet50, predicts MGMT promoter methylation status from the segmented volumes. The proposed framework demonstrated superior performance in predicting MGMT promoter methylation status in glioblastoma patients compared to conventional methods, achieving a classification accuracy of 95% and an AUC of 0.96. These results underscore the model's potential to enhance patient stratification and guide treatment selection. The accurate prediction of MGMT promoter methylation status via non-invasive imaging provides a reliable criterion for anticipating patient responsiveness to alkylating chemotherapy. This capability equips clinicians with a tool to inform personalized treatment strategies, optimizing therapeutic efficacy from the outset.