Back to all papers

Deep learning MGMT promoter methylation status prediction in IDH-wildtype glioblastoma using multiparametric MRI: Benchmark vs. real‑world performance.

November 16, 2026pubmed logopapers

Authors

Leone A,Filimonova E,Carbone F,Zoli M,Rzaev J,Schukina M,Carretta A,Bianconi A,Cofano F,Morello A,Di Perna G,Armocida D,Spetzger U,Roumia S,Di Napoli V,Fochi NP,Lau R,Internò V,Giordano G,Curcio A,Rustici A,Angileri F,Mazzatenta D,Signorelli F,Porta C,Garbossa D,Luu MSK,Benedichuk M,Shakoor A,Tuchinov B,Colamaria A

Affiliations (23)

  • Department of Neurosurgery, Städtisches Klinikum Karlsruhe, Karlsruher Neurozentrum, Karlsruhe, Germany.
  • Federal Neurosurgical Center, 132/1 Nemirovicha-Danchenko St, Novosibirsk, 630087, Russia.
  • The Artificial Intelligence Research Center of Novosibirsk State University, 1 Pirogov St, Novosibirsk, 630090, Russia.
  • Department of Neurosurgery, Städtisches Klinikum Karlsruhe, Karlsruher Neurozentrum, Karlsruhe, Germany. [email protected].
  • Division of Neurosurgery, University Hospital of Foggia, Policlinico "Riuniti", Foggia, Italy. [email protected].
  • Department of Biomedical and Neuromotor Sciences (DIBINEM), "Alma Mater Studiorum" University of Bologna, Bologna, Italy.
  • Programma Neurochirurgia Ipofisi-Pituitary Unit, IRCCS Istituto delle Scienze Neurologiche di Bologna, Bologna, Italy.
  • Department of Neuroscience, Rehabilitation, Ophthalmology, Genetics, Maternal and Child Health (DINOGMI), University of Genova, Genova, 16132, Italy.
  • Department of Neurosurgery, IRCCS Ospedale Policlinico San Martino, Genova, Genoa, 16132, Italy.
  • Department of Neurosurgery, AOU Città della Salute e della Scienza, Turin, Italy.
  • Department of Neurosurgery, University of Turin, Turin, Italy.
  • Division of Neurosurgery, University Hospital of Foggia, Policlinico "Riuniti", Foggia, Italy.
  • Institute for Anthropomatics and Robotics, Karlsruhe Institute for Technology - KIT, Karlsruhe, Germany.
  • Department of Radiology, Division of Neuroradiology, Städtisches Klinikum Karlsruhe, Karlsruhe, Germany.
  • Neurosurgery Department, Hospital Universitari Joan XXIII, Tarragona, Spain.
  • Universitat Rovira i Virgili, Tarragona, Spain.
  • Division of Medical Oncology, A.O.U. Consorziale Policlinico Di Bari, Bari, Italy.
  • Unit of Medical Oncology and Biomolecular Therapy and CREATE Center for Research and Innovation Medicine, Department of Medical and Surgical Sciences, University of Foggia, Policlinico Riuniti, Foggia, Italy.
  • UOC di Neurochirurgia, Fondazione IRCCS Casa Sollievo della Sofferenza, San Giovanni Rotondo, Italy.
  • Neuroradiology Unit, Ospedale Maggiore, IRCCS Istituto delle Scienze Neurologiche di Bologna, Bologna, Italy.
  • Neurosurgery Unit, University Hospital G. Martino, Messina, Italy.
  • Division of Neurosurgery, Department of Translational Biomedicine and Neurosciences, University Aldo Moro of Bari, Bari, Italy.
  • Interdisciplinary Department of Medicine, University of Bari Aldo Moro, Bari, Italy.

Abstract

Preoperative determination of O6-methylguanine-DNA methyltransferase (MGMT) promoter methylation status is of major prognostic and predictive relevance in IDH-wildtype glioblastoma, yet currently depends on invasive tissue sampling. This study investigates deep learning-based strategies for MGMT prediction from multiparametric MRIs, with particular emphasis on robustness and generalizability under heterogeneous, multicenter conditions. A total of 949 glioblastoma cases were analyzed, integrating public benchmark data from the Brain Tumor Segmentation Challenge (BraTS) with multi-institutional private cohorts acquired across diverse scanners and imaging protocols. Four complementary experimental settings were designed to compare reproduced literature baselines, private-only training, whole-brain end-to-end models, and segmentation-guided region-of-interest (ROI) pipelines. Tumor delineation was performed using an nnU-Net framework with expert quality control, and multiple three-dimensional convolutional neural network architectures were evaluated. Model performance was highest on the homogeneous BraTS benchmark, where a whole-brain DenseNet121 achieved an area under the receiver operating characteristic curve of 0.80. However, marked performance deterioration was observed when models were evaluated on independent held-out multicenter test set. In contrast, segmentation-guided ROI-based approaches showed a more stable performance profile in this preliminary independent held-out multicenter test. Explicit lesion localization was associated with reduction in false-positive classifications, suggesting a potential benefit for performance stability across heterogeneous datasets, although validation in external cohorts from institutions not represented during training is required. Collectively, these results demonstrate that upstream methodological choices, especially tumor segmentation and ROI definition, remain critical determinants of performance in MRI-based radiogenomic modeling. Future studies should emphasize prospective multicenter validation, domain-shift harmonization, multimodal data integration, and uncertainty-aware deep learning to enhance robustness and interpretability.

Topics

GlioblastomaDeep LearningBrain NeoplasmsDNA Repair EnzymesDNA Modification MethylasesTumor Suppressor ProteinsDNA MethylationMultiparametric Magnetic Resonance ImagingJournal Article

Ready to Sharpen Your Edge?

Subscribe to join 11k+ peers who rely on RadAISlice. Get the essential weekly briefing that empowers you to navigate the future of radiology.

We respect your privacy. Unsubscribe at any time.