Radiomics in Medulloblastoma: Advancing Non-Invasive Diagnosis, Molecular Subtyping, and Prognostication

Motivation: As medulloblastoma (MB) management advances, radiomics provides a non-invasive avenue to improve diagnosis, subtyping, and prognostication. By consolidating current evidence, this study clarifies radiomics' capability to address these critical aspects effectively. Goal(s): To systematically evaluate radiomics' accuracy in MB diagnosis, subtype differentiation, and prognostic insights, identifying specific imaging markers that could strengthen clinical decision-making. Approach: We analyzed 21 qualified studies to assess the efficacy of radiomic features. Results: Radiomics showed encouraging classification accuracies, with some models reaching up to 95% in differentiating MB fromothertumors. CE-T1, GLCM, and ADC metrics provided valuable insights into subtype characteristics and potential prognostic indicators. Impact: This study underscores radiomics' potential to refine non-invasive MB management, aiding clinicians in earlier, more precise diagnoses and individualized treatment plans. It highlights encouraging imaging markers that may inspire future research on radiomics-driven clinical protocols and patient outcomes.