Beyin tümörlerinde radyocerrahi plani kalitesinin değerlendirilmesi ve genetik mutasyon tespiti için manyetik rezonans görüntüleme'nin radyomik analizi

Magnetic resonance imaging (MRI) is essential in clinical practice for understanding brain tumor biology, optimizing treatment strategies, and predicting prognosis. Recent advancements in computation and artificial intelligence have integrated MRI into big data for precision medicine. Radiomics, a noninvasive and quantitative technique, converts imaging data into high-dimensional datasets, revealing encrypted information linking to the tumor pathophysiology to improve the precision of diagnosis, prognosis, and treatment planning. This thesis focused on specific brain tumors—vestibular schwannomas, gliomas, and meningiomas— and included clinical, anatomical imaging, and pathological data. We aimed to develop radiomics-based pipelines to: i) evaluate Gamma Knife dose planning quality and treatment outcomes, ii) predict isocitrate dehydrogenase (IDH) mutation status and prognosis in glioma patients with validation on an external dataset, and iii) assess alterations in neurofibromatosis type 2 (NF2) gene and predict recurrence-free survival (RFS) in meningiomas. The key findings of the dissertation included: tumor shape irregularity affected Gamma Knife dose planning quality without having an influence on treatment outcomes for vestibular schwannomas; the degree of sphericity and intensity patterns of gliomas on MRI were important predictors of IDH mutation status and prognosis; and structural MRI radiomics were strongly correlated with NF2 loss and RFS in meningiomas. These results highlighted the potential of radiomics in the decision-making framework, providing new insights into neurooncology.