Radiogenomics: A Machine Learning Augmentation to MRI-Based Glioma Profiling

In response to the article by Ganesan et al, on the role of magnetic resonance imaging (MRI) in differentiating glioma subtypes published in the Asian Journal of Neurosurgery.[1] The study highlights the potential of noninvasive MRI features such as T2- fluid-attenuated inversion recovery mismatch and necrosis quantification in distinguishing isocitrate dehydrogenase (IDH)-mutant from IDH-wild-type gliomas.

Building on these findings, we would like to extend the discussion toward the integration of machine learning (ML) approaches for genomic profiling of gliomas. Such integration of imaging with genomic and transcriptomic data, referred to as “radiogenomics,” holds promise in addressing glioma heterogeneity. Given the challenges of repeated biopsies and sampling bias in glioma diagnosis, ML-based radiogenomic models can noninvasively infer tumor biology, immune microenvironment, and even response to treatment.

Emerging literature has demonstrated the ability of ML models using MRI-derived radiological profiles, often referred to as radiomic feature sets, to predict molecular subtypes such as IDH mutation and deoxyribonucleic acid methylation profiles using two novel approaches—radiomic oncology (RO) and radiomic set enrichment analysis (RSEA).[2] RO employs a hypergeometric test to evaluate sets of radiomic imaging features associated with tumor biology, while RSEA—modeled after the Gene Set Enrichment Analysis (GSEA) framework[3]—leverages full radiomic feature sets to identify molecularly enriched imaging patterns. These methods were further advanced by Ji et al,[4] who combined MRI with molecular messenger ribonucleic acid data to classify IDH-wild-type glioblastomas into subtypes with distinct prognoses and therapy responses.

Together, these advances suggest a paradigm shift: from conventional radiology to radiogenomic intelligence. ML models can generate real-time genomic predictions from MRI scans, especially in resource-limited or biopsy-constrained settings. The clinical implications of radiogenomics include the possibility of noninvasive, low-cost genomic assessment, and the ability to capture spatially comprehensive tumor information—unlike conventional genome sequencing, which is often limited to a small biopsy specimen.[5] As the article by Ganesan et al rightly emphasizes imaging's potential in glioma evaluation, it becomes imperative to leverage artificial intelligence-driven radiogenomics into clinical workflows to reduce diagnostic delay and individualize glioma management.

Authors' Contributions

A.C.D. conceptualized the study, designed the manuscript structure, conducted the primary literature review, and drafted the initial version. A.D. assisted with data collection, language refinement, and final proofreading.

Publication History

Article published online:
13 August 2025

• References

• 1 Ganesan G, Rangasami R, Chandrasekharan A, Marreddy S, Ramachandran R. Role of advanced magnetic resonance imaging in differentiating among glioma subtypes and predicting tumor-proliferative behavior. Asian J Neurosurg 2024; 20 (01) 34-42
• 2 Zhuang Z, Lin J, Wan Z. et al. Radiogenomic profiling of global DNA methylation associated with molecular phenotypes and immune features in glioma. BMC Med 2024; 22 (01) 352
• 3 Subramanian A, Tamayo P, Mootha VK. et al. Gene Set Enrichment Analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 2005; 102 (43) 15545-15550
• 4 Ji Q, Zheng Y, Zhou L, Chen F, Li W. Unveiling divergent treatment prognoses in IDHwt-GBM subtypes through multiomics clustering: a swift dual MRI-mRNA model for precise subtype prediction. J Transl Med 2024; 22 (01) 578
• 5 Shui L, Ren H, Yang X. et al. The era of radiogenomics in precision medicine: an emerging approach to support diagnosis, treatment decisions, and prognostication in oncology. Front Oncol 2021; 10: 570465