Analysis of radiomic features and clinical data in glioblastoma patients

Background: Glioblastoma (GBM) is the most common and aggressive primary brain tumor of the central nervous system, with a median overall survival of approximately 15 months. This study aims to develop a predictive model for overall survival (OS) and progression-free survival (PFS) in GBM patients by integrating radiomic features extracted from computed tomography (CT) images acquired during radiotherapy simulation with clinical and molecular data using deep learning algorithms. Methods: The study included 115 GBM patients treated at Akdeniz University between 2018 and 2024. Clinical data were retrospectively collected. Radiomic features were extracted from CT images, and preprocessing steps were applied for standardization. Survival analyses were performed using Kaplan-Meier estimation and Cox regression models. Artificial intelligence-based machine learning algorithms were employed for model development. Results: The median age of the patients included in the study was 60 years, with a median OS of 13 months and a median PFS of 6 months. The one-year OS rate was determined to be 52.7%, while the six-month PFS rate was 62.6%. Factors negatively affecting survival included the presence of multifocal tumors, bilateral tumor involvement, poor performance status, and advanced age. Gross total resection, in conjunction with concurrent temozolomide therapy, was found to be significantly associated with enhanced survival outcomes. Among the predictive models, logistic regression demonstrated the highest accuracy (73.9%) for one-year OS prediction, whereas the gradient boosting model achieved the best performance (66.7% accuracy) for six-month PFS prediction. Conclusion: Integrating radiomic features with clinical and molecular data holds substantial potential for personalized prognostic modeling in GBM. These findings indicate that radiomics could be a valuable tool for enhancing treatment strategies. However, additional large-scale studies are necessary to validate these results.