{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,26]],"date-time":"2026-08-26T05:20:21Z","timestamp":1787721621236,"version":"build-2784847793"},"reference-count":35,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,8,26]],"date-time":"2026-08-26T00:00:00Z","timestamp":1787702400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Oncol."],"abstract":"<jats:sec>\n                    <jats:title>Introduction<\/jats:title>\n                    <jats:p>To develop and externally validate an explainable 3D convolutional neural network (CNN) ensemble for non-invasive prediction of IDH mutation and MGMT promoter methylation status in gliomas using routine MRI.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>Multi-institutional MRI data were curated from the UCSF-PDGM and UPENN-GBM cohorts for training and testing, with independent external validation on the MU-Glioma-Post dataset. Separate 3D CNNs were trained for IDH and MGMT classification, and the models were combined into an ensemble metaclassifier. Model interpretability was assessed using Integrated Gradients saliency maps for voxel-level attributions and Shapley values for modality-level contributions.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>The ensemble model achieved an AUC of 0.94 (95% CI, 0.84\u20131.00) in testing and 0.77 (95% CI, 0.71\u20130.82) in external validation for IDH mutation classification. The ensemble yielded AUCs of 0.81 (95% CI, 0.64\u20130.94), 0.68 (95% CI, 0.47\u20130.88), and 0.57 (95% CI, 0.52\u20130.63) for MGMT promoter methylation in validation, testing, and external validation, respectively. Integrated Gradients (IG) maps highlighted biologically plausible regions, with IDH predictions focusing on peritumoral FLAIR hyperintensity and MGMT predictions emphasizing the enhancing tumor core on T1CE. Shapley analysis showed FLAIR contributed 64% and T1CE 36% to IDH classification, while MGMT classification relied more heavily on T1CE (60.6%), with complementary input from FLAIR (30.8%).<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Discussion<\/jats:title>\n                    <jats:p>We present an explainable 3D ensemble deep learning framework capable of predicting IDH mutation status from routine MRI with moderate external generalizability. While the same framework achieved strong internal performance for MGMT methylation status, this did not generalize on external validation, underscoring that MGMT prediction from imaging alone remains an unresolved challenge requiring further methodological development.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fonc.2026.1790776","type":"journal-article","created":{"date-parts":[[2026,8,26]],"date-time":"2026-08-26T04:42:26Z","timestamp":1787719346000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Externally validated explainable 3D CNN ensemble model for non-invasive prediction of IDH and MGMT status in gliomas"],"prefix":"10.3389","volume":"16","author":[{"given":"Rafail C.","family":"Christodoulou","sequence":"first","affiliation":[{"name":"Department of Radiology, Stanford University School of Medicine","place":["Stanford, CA, United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Georgios","family":"Vamvouras","sequence":"additional","affiliation":[{"name":"Department of Mechanical Engineering, National Technical University of Athens","place":["Zografou, Greece"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Rafael","family":"Pitsillos","sequence":"additional","affiliation":[{"name":"Department of Bioinformatics, Cyprus Institute of Neurology and Genetics","place":["Nicosia, Cyprus"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Kleoniki","family":"Roka","sequence":"additional","affiliation":[{"name":"Department of Pediatric Oncology, University of Athens Medical School","place":["Athens, Greece"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Elena E.","family":"Solomou","sequence":"additional","affiliation":[{"name":"Department of Internal Medicine-Hematology, University of Patras Medical School","place":["Rion, Greece"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Christina","family":"Kalogeropoulou","sequence":"additional","affiliation":[{"name":"Department of Radiology, University of Patras Medical School, General Hospital of Patras","place":["Patras, Greece"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Peter","family":"Zampakis","sequence":"additional","affiliation":[{"name":"Department of Radiology, University of Patras Medical School, General Hospital of Patras","place":["Patras, Greece"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Michalis F.","family":"Georgiou","sequence":"additional","affiliation":[{"name":"Department of Radiology, Division of Nuclear Medicine, University of Miami","place":["Miami, FL, United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,8,26]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","DOI":"10.1093\/neuonc\/noae145","article-title":"CBTRUS statistical report: Primary brain and other central nervous system tumors diagnosed in the United States in 2017\u20132021","volume":"26","author":"Price","year":"2024","journal-title":"Neuro-Oncology"},{"key":"B2","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-021-89477-w","article-title":"Generative adversarial network for glioblastoma ensures morphologic variations and improves diagnostic model for isocitrate dehydrogenase mutant type","volume":"11","author":"Park","year":"2021","journal-title":"Sci Rep"},{"key":"B3","doi-asserted-by":"publisher","first-page":"997","DOI":"10.1056\/NEJMoa043331","article-title":"MGMT gene silencing and benefit from temozolomide in glioblastoma","volume":"352","author":"Hegi","year":"2005","journal-title":"N Engl J Med"},{"key":"B4","doi-asserted-by":"publisher","DOI":"10.1007\/s00330-018-5608-7","article-title":"Imaging prediction of isocitrate dehydrogenase (IDH) mutation in patients with glioma: a systemic review and meta-analysis","volume":"29","author":"Suh","year":"2019","journal-title":"Eur Radiol"},{"key":"B5","doi-asserted-by":"publisher","DOI":"10.1093\/neuonc\/nov095","article-title":"Consensus recommendations for a standardized brain tumor imaging protocol in clinical trials","volume":"17","author":"Ellingson","year":"2015","journal-title":"Neuro Oncol"},{"key":"B6","doi-asserted-by":"publisher","DOI":"10.1002\/jmri.28743","article-title":"The 2021 WHO classification for gliomas and implications on imaging diagnosis: Part 1\u2014Key points of the fifth edition and summary of imaging findings on adult-type diffuse gliomas","volume":"58","author":"Park","year":"2023","journal-title":"J Magn Reson Imaging"},{"key":"B7","doi-asserted-by":"publisher","DOI":"10.1016\/j.jmir.2019.09.005","article-title":"Machine learning and deep learning in medical imaging: Intelligent imaging","volume":"50","author":"Currie","year":"2019","journal-title":"J Med Imaging Radiat Sci"},{"key":"B8","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2024.103435","article-title":"Cooperative multi-task learning and interpretable image biomarkers for glioma grading and molecular subtyping","volume":"101","author":"Chen","year":"2025","journal-title":"Med Image Anal"},{"key":"B9","doi-asserted-by":"publisher","DOI":"10.1016\/j.compmedimag.2020.101831","article-title":"Automated MRI based pipeline for segmentation and prediction of grade, IDH mutation and 1p19q co-deletion in glioma","volume":"88","author":"Decuyper","year":"2021","journal-title":"Comput Med Imaging Graphics"},{"key":"B10","doi-asserted-by":"publisher","first-page":"4448","DOI":"10.1038\/s41598-025-87127-z","article-title":"A deep ensemble learning framework for glioma segmentation and grading prediction","volume":"15","author":"Wen","year":"2025","journal-title":"Sci Rep"},{"key":"B11","doi-asserted-by":"publisher","DOI":"10.7937\/tcia.bdgf-8v37","article-title":"The university of California San Francisco preoperative diffuse glioma MRI (UCSF-PDGM) (Version 5) [dataset","author":"Calabrese","year":"2022","journal-title":"Cancer Imaging Arch"},{"key":"B12","doi-asserted-by":"publisher","DOI":"10.7937\/TCIA.709X-DN49","article-title":"Multi-parametric magnetic resonance imaging (mpMRI) scans for de novo Glioblastoma (GBM) patients from the University of Pennsylvania Health System (UPENN-GBM) (Version 2) [Data set","author":"Bakas","year":"2021","journal-title":"Cancer Imaging Arch"},{"key":"B13","doi-asserted-by":"publisher","DOI":"10.7937\/7K9K-3C83","article-title":"University of missouri post-operative glioma dataset (MU-glioma-post) (Version 1) [Data set","author":"Yaseen","year":"2025","journal-title":"Cancer Imaging Arch"},{"key":"B14","doi-asserted-by":"publisher","DOI":"10.1002\/cncr.27392","article-title":"Comparative assessment of 5 methods (methylation-specific polymerase chain reaction, methylight, pyrosequencing, methylation-sensitive high-resolution melting, and immunohistochemistry) to analyze O6-methylguanine-DNA-methyltranferase in a series of 100 glioblastoma patients","volume":"118","author":"Quillien","year":"2012","journal-title":"Cancer"},{"key":"B15","doi-asserted-by":"publisher","first-page":"3273","DOI":"10.1038\/s41598-025-87803-0","article-title":"Deep learning classification of MGMT status of glioblastomas using multiparametric MRI with a novel domain knowledge augmented mask fusion approach","volume":"15","author":"Koska","year":"2025","journal-title":"Sci Rep"},{"key":"B16","unstructured":"Optuna: A next-generation hyperparameter optimization framework\n          \n          \n            \n              Akiba\n              T\n            \n            \n              Sano\n              S\n            \n            \n              Yanase\n              T\n            \n            \n              Ohta\n              T\n            \n            \n              Koyama\n              M\n            \n          \n          \n          2019"},{"key":"B17","doi-asserted-by":"publisher","first-page":"27533","DOI":"10.1038\/s41598-025-08608-9","article-title":"Prediction of MGMT methylation status in glioblastoma patients based on radiomics feature extracted from intratumoral and peritumoral MRI imaging","volume":"15","author":"Chen","year":"2025","journal-title":"Sci Rep"},{"key":"B18","doi-asserted-by":"publisher","DOI":"10.3390\/cancers15051384","article-title":"Peritumoral ADC values correlate with the MGMT methylation status in patients with glioblastoma","volume":"15","author":"Ladenhauf","year":"2023","journal-title":"Cancers"},{"key":"B19","doi-asserted-by":"publisher","DOI":"10.3390\/biology13110885","article-title":"IDH mutations in glioma: Molecular, cellular, diagnostic, and clinical implications","volume":"13","author":"Choate","year":"2024","journal-title":"Biology"},{"key":"B20","doi-asserted-by":"publisher","DOI":"10.1093\/neuonc\/noz095","article-title":"Prediction of IDH genotype in gliomas with dynamic susceptibility contrast perfusion MR imaging using an explainable recurrent neural network","volume":"21","author":"Choi","year":"2019","journal-title":"Neuro-Oncology"},{"key":"B21","doi-asserted-by":"publisher","DOI":"10.1093\/neuonc\/noaa177","article-title":"Fully automated hybrid approach to predict the IDH mutation status of gliomas via deep learning and radiomics","volume":"23","author":"Choi","year":"2021","journal-title":"Neuro-Oncology"},{"key":"B22","unstructured":"Diagnostic performance of deep learning for predicting glioma isocitrate dehydrogenase and 1p\/19q co-deletion in MRI: a systematic review and meta-analysis\n          \n          \n            \n              Farahani\n              S\n            \n            \n              Hejazi\n              M\n            \n            \n              Tabassum\n              M\n            \n            \n              Di Ieva\n              A\n            \n            \n              Mahdavifar\n              N\n            \n            \n              Liu\n              S\n            \n          \n          \n          2025"},{"key":"B23","doi-asserted-by":"publisher","first-page":"140","DOI":"10.1038\/s41746-025-01530-4","article-title":"Interpretable multimodal transformer for prediction of molecular subtypes and grades in adult-type diffuse gliomas","volume":"8","author":"Byeon","year":"2025","journal-title":"NPJ Digit Med"},{"key":"B24","doi-asserted-by":"publisher","first-page":"181","DOI":"10.1038\/s41698-024-00670-2","article-title":"Biologically interpretable multi-task deep learning pipeline predicts molecular alterations, grade, and prognosis in glioma patients","volume":"8","author":"Wu","year":"2024","journal-title":"NPJ Precis Onc"},{"key":"B25","doi-asserted-by":"publisher","DOI":"10.1016\/j.displa.2025.103072","article-title":"A systematic review and meta-analysis of deep learning and radiomics in predicting MGMT promoter methylation status in glioblastoma: Efficacy, reliability, and clinical implications","volume":"89","author":"Chen","year":"2025","journal-title":"Displays"},{"key":"B26","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2023.102989","article-title":"MGMT promoter methylation status prediction using MRI scans? An extensive experimental evaluation of deep learning models","volume":"90","author":"Saeed","year":"2023","journal-title":"Med Image Anal"},{"key":"B27","doi-asserted-by":"publisher","first-page":"589","DOI":"10.1056\/NEJMoa2304194","article-title":"Vorasidenib in IDH1- or IDH2-mutant low-grade glioma","volume":"389","author":"Mellinghoff","year":"2023","journal-title":"N Engl J Med"},{"key":"B28","doi-asserted-by":"publisher","first-page":"394","DOI":"10.1007\/s00701-024-06300-x","article-title":"MGMT methylation and its prognostic significance in inoperable IDH-wildtype glioblastoma: the MGMT-GBM study","volume":"166","author":"Ghimire","year":"2024","journal-title":"Acta Neurochir (Wien)"},{"key":"B29","doi-asserted-by":"publisher","DOI":"10.1016\/j.ejca.2020.06.018","article-title":"Histopathological grading affects survival in patients with IDH-mutant grade II and grade III diffuse gliomas","volume":"137","author":"Franceschi","year":"2020","journal-title":"Eur J Cancer"},{"key":"B30","doi-asserted-by":"publisher","DOI":"10.1016\/j.wneu.2018.01.045","article-title":"The prognostic roles of gender and O6-methylguanine-DNA methyltransferase methylation status in glioblastoma patients: The female power","volume":"112","author":"Franceschi","year":"2018","journal-title":"World Neurosurg"},{"key":"B31","doi-asserted-by":"publisher","first-page":"1847","DOI":"10.1038\/s41597-025-06011-7","article-title":"MU-Glioma Post: A comprehensive dataset of automated MR multi-sequence segmentation and clinical features","volume":"12","author":"Mahmoud","year":"2025","journal-title":"Sci Data"},{"key":"B32","doi-asserted-by":"publisher","DOI":"10.2353\/ajpath.2009.080958","article-title":"IDH1 mutations are early events in the development of astrocytomas and oligodendrogliomas","volume":"174","author":"Watanabe","year":"2009","journal-title":"Am J Pathol"},{"key":"B33","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-019-1775-1","article-title":"Longitudinal molecular trajectories of diffuse glioma in adults","volume":"576","author":"Barthel","year":"2019","journal-title":"Nature"},{"key":"B34","doi-asserted-by":"publisher","DOI":"10.1111\/neup.70047","article-title":"Dynamics of MGMT methylation status in Malignant gliomas with and without therapeutic intervention: Significance of two\u2010dimensional immunohistochemical analysis and evaluation of recurrent specimens","volume":"46","author":"Asano","year":"2026","journal-title":"Neuropathology"},{"key":"B35","doi-asserted-by":"publisher","DOI":"10.1093\/neuonc\/noaf298","article-title":"Epigenetic evolution of isocitrate dehydrogenase-wildtype glioblastomas","volume":"28","author":"Deng","year":"2026","journal-title":"Neuro Oncol"}],"container-title":["Frontiers in Oncology"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fonc.2026.1790776\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,26]],"date-time":"2026-08-26T04:42:33Z","timestamp":1787719353000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fonc.2026.1790776\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,26]]},"references-count":35,"alternative-id":["10.3389\/fonc.2026.1790776"],"URL":"https:\/\/doi.org\/10.3389\/fonc.2026.1790776","relation":{},"ISSN":["2234-943X"],"issn-type":[{"value":"2234-943X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,26]]},"article-number":"1790776"}}