{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,12]],"date-time":"2026-06-12T17:58:29Z","timestamp":1781287109200,"version":"3.54.1"},"reference-count":31,"publisher":"Public Library of Science (PLoS)","issue":"6","license":[{"start":{"date-parts":[[2026,6,12]],"date-time":"2026-06-12T00:00:00Z","timestamp":1781222400000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["www.plosone.org"],"crossmark-restriction":false},"short-container-title":["PLoS One"],"abstract":"<jats:p>Glioblastoma multiforme (GBM) is the most aggressive primary brain tumor in adults, with a median survival of 14.6 months under standard radiotherapy and temozolomide (TMZ) chemotherapy. The methylation status of the O\u2076-methylguanine-DNA methyltransferase (MGMT) promoter is a critical biomarker predicting TMZ response; however, its determination currently requires invasive tissue sampling. Non-invasive prediction of MGMT promoter methylation from multiparametric MRI (mpMRI) through deep learning represents a compelling alternative, yet its clinical feasibility remains unresolved. Using the BraTS 2021 dataset (582 patients, four MRI sequences: FLAIR, T1w, T1wCE, T2w), we conducted a systematic comparative study of unimodal and multimodal deep learning approaches based on VGG-16, exploring 1,380 experimental configurations (unimodal: 192; multimodal: 1,188) across three imaging planes, eight slice counts, and three multimodal fusion strategies (early, intermediate, and late fusion). In the unimodal setting, the best model trained on T2w coronal images (32 slices, no transfer learning) achieved an accuracy of 0.6458 and an AUC of 0.6422 on the validation set, but dropped to 0.5586 and 0.5533 on the independent test set, revealing substantial overfitting attributable to limited dataset size. Strikingly, multimodal fusion consistently failed to outperform the best unimodal model, with all three fusion strategies plateauing at ~0.64 accuracy and ~0.64 AUC on validation data. Transfer learning improved generalization across train\/test distributions at the cost of peak performance. These findings suggest, for the tested framework in this study, that MGMT methylation status prediction from mpMRI remains fundamentally constrained by dataset heterogeneity and size, irrespective of modality combination strategy, and that T2w coronal acquisitions could be more interesting in future data collection efforts.<\/jats:p>","DOI":"10.1371\/journal.pone.0351405","type":"journal-article","created":{"date-parts":[[2026,6,12]],"date-time":"2026-06-12T17:30:02Z","timestamp":1781285402000},"page":"e0351405","update-policy":"https:\/\/doi.org\/10.1371\/journal.pone.corrections_policy","source":"Crossref","is-referenced-by-count":0,"title":["Unimodal vs. multimodal deep learning for non-invasive MGMT promoter methylation prediction in glioblastoma: A systematic evaluation on the BraTS 2021 dataset"],"prefix":"10.1371","volume":"21","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-2497-6983","authenticated-orcid":true,"given":"Freddy","family":"Oulia","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Philippe","family":"Charton","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2488-1653","authenticated-orcid":true,"given":"Muhammad","family":"Kabir","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Fabrice","family":"Gardebien","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"C\u00e9dric","family":"Damour","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3568-9595","authenticated-orcid":true,"given":"Frederic","family":"Cadet","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2026,6,12]]},"reference":[{"issue":"1","key":"pone.0351405.ref001","doi-asserted-by":"crossref","first-page":"93","DOI":"10.1007\/s00401-005-0991-y","article-title":"Epidemiology and etiology of gliomas","volume":"109","author":"H Ohgaki","year":"2005","journal-title":"Acta Neuropathol"},{"issue":"8","key":"pone.0351405.ref002","doi-asserted-by":"crossref","first-page":"1231","DOI":"10.1093\/neuonc\/noab106","article-title":"The 2021 WHO Classification of Tumors of the Central Nervous System: A summary","volume":"23","author":"DN Louis","year":"2021","journal-title":"Neuro Oncol"},{"issue":"5","key":"pone.0351405.ref003","doi-asserted-by":"crossref","first-page":"492","DOI":"10.1056\/NEJMra0708126","article-title":"Malignant gliomas in adults","volume":"359","author":"PY Wen","year":"2008","journal-title":"N Engl J Med"},{"issue":"10","key":"pone.0351405.ref004","doi-asserted-by":"crossref","first-page":"987","DOI":"10.1056\/NEJMoa043330","article-title":"Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma","volume":"352","author":"R Stupp","year":"2005","journal-title":"N Engl J Med"},{"key":"pone.0351405.ref005","doi-asserted-by":"crossref","first-page":"462","DOI":"10.1016\/j.cell.2013.09.034","article-title":"The somatic genomic landscape of glioblastoma","volume":"155","author":"CW Brennan","year":"2013","journal-title":"Cell"},{"key":"pone.0351405.ref006","doi-asserted-by":"crossref","DOI":"10.15586\/codon.glioblastoma.2017.ch11","article-title":"Current standards of care in glioblastoma therapy","volume-title":"Glioblastoma","author":"C Fernandes","year":"2017"},{"issue":"1","key":"pone.0351405.ref007","doi-asserted-by":"crossref","DOI":"10.1371\/journal.pone.0016146","article-title":"Comprehensive analysis of MGMT promoter methylation: Correlation with MGMT expression and clinical response in GBM","volume":"6","author":"N Shah","year":"2011","journal-title":"PLoS One"},{"key":"pone.0351405.ref008","doi-asserted-by":"crossref","first-page":"997","DOI":"10.1056\/NEJMoa043331","article-title":"MGMT gene silencing and benefit from temozolomide in glioblastoma","volume":"352","author":"ME Hegi","year":"2005","journal-title":"New England Journal of Medicine"},{"issue":"83","key":"pone.0351405.ref009","first-page":"221","article-title":"Advances and challenges in the treatment of glioblastoma: A clinician\u2019s perspective","volume":"15","author":"MM Mrugala","year":"2013","journal-title":"Discov Med"},{"issue":"1","key":"pone.0351405.ref010","doi-asserted-by":"crossref","first-page":"39","DOI":"10.1038\/nrneurol.2009.197","article-title":"MGMT promoter methylation in malignant gliomas: Ready for personalized medicine?","volume":"6","author":"M Weller","year":"2010","journal-title":"Nat Rev Neurol"},{"issue":"5","key":"pone.0351405.ref011","doi-asserted-by":"crossref","first-page":"380","DOI":"10.1016\/j.trecan.2020.02.010","article-title":"MGMT Status as a Clinical Biomarker in Glioblastoma","volume":"6","author":"M Butler","year":"2020","journal-title":"Trends Cancer"},{"issue":"10","key":"pone.0351405.ref012","doi-asserted-by":"crossref","first-page":"1213","DOI":"10.1007\/s11912-024-01580-z","article-title":"Radiogenomics as an integrated approach to glioblastoma precision medicine","volume":"26","author":"I Sanchez","year":"2024","journal-title":"Curr Oncol Rep"},{"issue":"6","key":"pone.0351405.ref013","doi-asserted-by":"crossref","first-page":"428","DOI":"10.1038\/s41571-024-00891-1","article-title":"Multiparametric MRI for characterization of the tumour microenvironment","volume":"21","author":"E Hoffmann","year":"2024","journal-title":"Nat Rev Clin Oncol"},{"key":"pone.0351405.ref014","article-title":"The RSNA-ASNR-MICCAI BraTS 2021 benchmark on brain tumor segmentation and radiogenomic classification","author":"U Baid","year":"2021"},{"issue":"10","key":"pone.0351405.ref015","doi-asserted-by":"crossref","first-page":"1993","DOI":"10.1109\/TMI.2014.2377694","article-title":"The multimodal brain tumor image segmentation benchmark (BRATS)","volume":"34","author":"BH Menze","year":"2015","journal-title":"IEEE Trans Med Imaging"},{"key":"pone.0351405.ref016","doi-asserted-by":"crossref","first-page":"170117","DOI":"10.1038\/sdata.2017.117","article-title":"Advancing The Cancer Genome Atlas glioma MRI collections with expert segmentation labels and radiomic features","volume":"4","author":"S Bakas","year":"2017","journal-title":"Sci Data"},{"key":"pone.0351405.ref017","author":"A Dosovitskiy","year":"2020"},{"key":"pone.0351405.ref018","author":"Z Liu","year":"2021"},{"issue":"8","key":"pone.0351405.ref019","doi-asserted-by":"crossref","first-page":"2253","DOI":"10.3390\/cancers15082253","article-title":"MRI-Based Deep Learning Tools for MGMT promoter methylation detection: A thorough evaluation","volume":"15","author":"L Robinet","year":"2023","journal-title":"Cancers (Basel)"},{"key":"pone.0351405.ref020","doi-asserted-by":"crossref","first-page":"102989","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":"N Saeed","year":"2023","journal-title":"Med Image Anal"},{"issue":"3","key":"pone.0351405.ref021","doi-asserted-by":"crossref","first-page":"124","DOI":"10.3390\/info13030124","article-title":"An attentive multi-modal CNN for brain tumor radiogenomic classification","volume":"13","author":"R Qu","year":"2022","journal-title":"Information"},{"issue":"6","key":"pone.0351405.ref022","doi-asserted-by":"crossref","first-page":"1045","DOI":"10.1007\/s10278-013-9622-7","article-title":"The Cancer Imaging Archive (TCIA): Maintaining and operating a public information repository","volume":"26","author":"K Clark","year":"2013","journal-title":"J Digit Imaging"},{"key":"pone.0351405.ref023","article-title":"The Cancer Genome Atlas (TCGA): An immeasurable source of knowledge","volume":"19","author":"K Tomczak","year":"2015","journal-title":"Contemp Oncol (Pozn)"},{"key":"pone.0351405.ref024","doi-asserted-by":"crossref","first-page":"1201","DOI":"10.3174\/ajnr.A5667","article-title":"Deep-learning convolutional neural networks accurately classify genetic mutations in gliomas","volume":"39","author":"P Chang","year":"2018","journal-title":"AJNR Am J Neuroradiol"},{"issue":"5","key":"pone.0351405.ref025","doi-asserted-by":"crossref","first-page":"845","DOI":"10.3174\/ajnr.A7029","article-title":"MRI-Based Deep-Learning Method for Determining Glioma MGMT Promoter Methylation Status","volume":"42","author":"CGB Yogananda","year":"2021","journal-title":"AJNR Am J Neuroradiol"},{"issue":"3","key":"pone.0351405.ref026","doi-asserted-by":"crossref","first-page":"251","DOI":"10.3390\/diagnostics15030251","article-title":"Virtual biopsy for the prediction of MGMT promoter methylation in Gliomas: A comprehensive review of radiomics and deep learning approaches applied to MRI","volume":"15","author":"A Leone","year":"2025","journal-title":"Diagnostics (Basel)"},{"issue":"3","key":"pone.0351405.ref027","doi-asserted-by":"crossref","first-page":"703","DOI":"10.1007\/s13246-021-01019-w","article-title":"Improved U-Net architecture with VGG-16 for brain tumor segmentation","volume":"44","author":"S Ghosh","year":"2021","journal-title":"Phys Eng Sci Med"},{"key":"pone.0351405.ref028","doi-asserted-by":"crossref","first-page":"1424546","DOI":"10.3389\/fonc.2024.1424546","article-title":"Precision lung cancer screening from CT scans using a VGG16-based convolutional neural network","volume":"14","author":"H Xu","year":"2024","journal-title":"Front Oncol"},{"issue":"19","key":"pone.0351405.ref029","doi-asserted-by":"crossref","first-page":"4827","DOI":"10.3390\/cancers14194827","article-title":"Validation of MRI-based models to predict MGMT promoter methylation in Gliomas: BraTS 2021 radiogenomics challenge","volume":"14","author":"B-H Kim","year":"2022","journal-title":"Cancers (Basel)"},{"issue":"1","key":"pone.0351405.ref030","doi-asserted-by":"crossref","first-page":"172","DOI":"10.1186\/s40644-024-00817-1","article-title":"Assessment of MGMT promoter methylation status in glioblastoma using deep learning features from multi-sequence MRI of intratumoral and peritumoral regions","volume":"24","author":"X Yu","year":"2024","journal-title":"Cancer Imaging"},{"issue":"1","key":"pone.0351405.ref031","doi-asserted-by":"crossref","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":"\u0130\u00d6 Koska","year":"2025","journal-title":"Sci Rep"}],"container-title":["PLOS One"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0351405","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,12]],"date-time":"2026-06-12T17:30:22Z","timestamp":1781285422000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0351405"}},"subtitle":[],"editor":[{"given":"Michael C","family":"Burger","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,6,12]]},"references-count":31,"journal-issue":{"issue":"6","published-online":{"date-parts":[[2026,6,12]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pone.0351405","relation":{},"ISSN":["1932-6203"],"issn-type":[{"value":"1932-6203","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,6,12]]}}}