Fig. 4

Clinical utility of the radiomics signature. The decision curve of the radiomics signature in the primary cohort (a). The x axis represented the threshold probability, where the expected benefit of treatment as MGMT methylated is equal to the expected benefit of treatment as MGMT unmethylated (the threshold probability varies from patient to patient). The y axis indicated the net benefit for the treatment which considered the benefit of true positive and loss of false positive, and higher net benefit value indicates better model. The net benefit of the radiomics signature is further compared with the default strategies, which we treat all patients as MGMT methylated (red line) or as MGMT unmethylated (black line). The current prediction model outweigh both default strategies at any threshold probability, suggesting the clinical value of our model at all circumstances. Kaplan-Meier curves revealed the prognosis-based groups stratified by the MGMT promoter methylation status and the radiomics signature (b)