Imaging Radiomics Features for Recurrence Risk in Glioma Treatment Pathways

Abstract

Gliomas represent the most prevalent and aggressive primary brain tumors in adults, characterized by a notoriously high rate of recurrence despite advancements in multimodal therapeutic interventions. The inherent intratumoral heterogeneity and complex microenvironment of gliomas necessitate advanced diagnostic and prognostic tools to tailor post-operative and adjuvant treatment pathways effectively. This paper investigates the utility of imaging radiomics features, coupled with a novel mixed evaluation framework, to predict recurrence risk in glioma patients. By extracting high-dimensional quantitative data from standard-of-care magnetic resonance imaging, radiomics provides a non-invasive surrogate for genomic and transcriptomic profiling. The proposed mixed evaluation paradigm integrates these radiomic signatures with foundational clinical and histomolecular parameters, creating a multidimensional prognostic matrix. Through rigorous feature engineering and predictive modeling, we analyze the spatial and textual heterogeneity of tumor sub-regions and their correlation with disease-free survival intervals. The results demonstrate that incorporating mixed evaluation methodologies significantly enhances the stratification of patients into high and low recurrence risk categories, thereby outperforming traditional clinical prognostic models. This comprehensive analysis emphasizes the potential of radiomics as an indispensable component of precision oncology, ultimately facilitating dynamic, personalized treatment modifications and improving overall clinical workflows in neuro-oncology.