Share：
Share this content in WeChat
X
[Chinese] [PDF] 11 1
• Clinical Article •
Deep neural network MRI radiomics predicts glioblastoma MGMT promoter methylation status
LÜ Chong  XIA Linfeng  CHEN Qingshui  ZHENG Guangxin  HUANG Biyun  CHEN Lei 

Cite this article as: LÜ C, XIA L F, CHEN Q S, et al. Deep neural network MRI radiomics predicts glioblastoma MGMT promoter methylation status[J]. Chin J Magn Reson Imaging, 2025, 16(10): 35-40. DOI:10.12015/issn.1674-8034.2025.10.006.


[Abstract] Objective To explore the value of a deep neural network model based on multi-sequence MRI in predicting the methylation status of the O6-methylguanine-DNA methyltransferase (MGMT) promoter in patients with glioblastoma.Materials and Methods T1WI and contrast enhanced T1-weighted imaging (CE-T1WI) data from 262 glioblastoma patients (162 methylation and 100 unmethylation) were retrospectively analyzed. The Mann-Whitney U test, least absolute shrinkage and selection operator (LASSO) regression analysis, combined with the Pearson correlation coefficient method, were used to screen the features. Based on the screened features, a prediction model was constructed by means of the deep neural network algorithm. To evaluate the prediction efficiency of this model, the area under the receiver operating characteristic curve (AUC) was adopted to measure the prediction accuracy and reliability of the model.Results The T1WI model (AUC = 0.752 in the validation set, sensitivity = 68.8%, specificity = 75.0%), the CE-T1WI model (AUC = 0.823 in the validation set, sensitivity = 75.0%, specificity = 75.0%), and the multi sequence combined model (AUC = 0.847 in the validation set, sensitivity = 81.3%, specificity = 80.0%) based on the deep neural network could be used to predict the MGMT promoter methylation of patients with glioblastoma, and the multi sequence combined model had the highest diagnostic efficacy compared with the single sequence models.Conclusions The multi sequence MRI radiomics model based on deep neural network can noninvasively predict the MGMT promoter methylation features in glioblastomas.
[Keywords] glioblastomas；multi-sequence magnetic resonance images；radiomics；deep neural network；O6-methylguanine-DNA methyltransferase

LÜ Chong   XIA Linfeng   CHEN Qingshui   ZHENG Guangxin   HUANG Biyun   CHEN Lei*  

Department of Radiology, Xiamen Third Hospital, Xiamen 361000, China.

Corresponding author: CHEN L, E-mail: 41426393@qq.com

Conflicts of interest   None.

Received  2025-05-05
Accepted  2025-09-03
DOI: 10.12015/issn.1674-8034.2025.10.006
Cite this article as: LÜ C, XIA L F, CHEN Q S, et al. Deep neural network MRI radiomics predicts glioblastoma MGMT promoter methylation status[J]. Chin J Magn Reson Imaging, 2025, 16(10): 35-40. DOI:10.12015/issn.1674-8034.2025.10.006.

[1]
WELLER M, WEN P Y, CHANG S M, et al. Glioma[J/OL]. Nat Rev Dis Primers, 2024, 10: 33 [2025-05-05]. https://www.nature.com/articles/s41572-024-00516-y. DOI: 10.1038/s41572-024-00516-y.
[2]
SCHAFF L R, MELLINGHOFF I K. Glioblastoma and Other Primary Brain Malignancies in Adults: A Review[J]. JAMA, 2023, 329: 574-587. DOI: 10.1001/jama.2023.0023.
[3]
MONTELLA L, CUOMO M, DEL GAUDIO N, et al. Epigenetic alterations in glioblastomas: Diagnostic, prognostic and therapeutic relevance[J]. Int J Cancer, 2023, 153: 476-488. DOI: 10.1002/ijc.34381.
[4]
CASTRESANA J S, MELENDEZ B. Glioblastoma biology, genetics and possible therapies[J/OL]. Cells, 2023, 12(16): 2063 [2025-05-05]. https://pmc.ncbi.nlm.nih.gov/articles/PMC10453586/. DOI: 10.3390/cells12162063.
[5]
PUHRINGER K, CZARDA P, ILUCA S, et al. Association of intergenic and intragenic MGMT enhancer methylation with MGMT promoter methylation, MGMT protein expression and clinical and demographic parameters in glioblastoma[J/OL]. Int J Mol Sci, 2025, 26(7): 3390 [2025-05-05]. https://pmc.ncbi.nlm.nih.gov/articles/PMC11990025/. DOI: 10.3390/ijms26073390.
[6]
LAM K, ELDRED B S C, KEVAN B, et al. Prognostic value of O6-methylguanine-DNA methyltransferase methylation in isocitrate dehydrogenase mutant gliomas[J/OL]. Neurooncol Adv, 2022, 4(1): vdac030 [2025-05-05]. https://pmc.ncbi.nlm.nih.gov/articles/PMC8982195/. DOI: 10.1093/noajnl/vdac030.
[7]
DE GODOY L L, RAJAN A, BANIHASHEMI A, et al. Response assessment in long-term glioblastoma survivors using a multiparametric MRI-based prediction model[J/OL]. Brain Sci, 2025, 15(2): 146 [2025-05-05]. https://pmc.ncbi.nlm.nih.gov/articles/PMC11852837/. DOI: 10.3390/brainsci15020146.
[8]
ZHANG Y R, HUA Y F, ZHU X Y, et al. Research progress of MRI-based texture analysis in high-grade gliomas[J]. Chin J Magn Reson Imaging, 2023, 14(2): 174-178. DOI: 10.12015/issn.1674-8034.2023.02.031.
[9]
CHEN D, ZHANG S M, YANG X B, et al. Research Progress on MRI in Neural Remodeling of Gliomas[J]. International Journal of Medical Radiology, 2024, 47: 390-395. DOI: 10.19300/j.2024.Z21472.
[10]
KWIAKOWSKA-MIERNIK A, MRUK B, SKLINDA K, et al. Radiomics in the diagnosis of glioblastoma[J/OL]. Pol J Radiol, 2023, 88: e461-e466 [2025-05-05]. https://pmc.ncbi.nlm.nih.gov/articles/PMC10660137/. DOI: 10.5114/pjr.2023.132168.
[11]
BIJARI S, REZAEIJO S M, SAYFOOLLAHI S, et al. Development and validation of a robust MRI-based nomogram incorporating radiomics and deep features for preoperative glioma grading: a multi-center study[J]. Quant Imaging Med Surg, 2025, 15(2): 1125-1138. DOI: 10.21037/qims-24-1543.
[12]
LI D, WANG X C. Research progress of magnetic resonance radiomics in predicting the methylation status of MGMT promoter in glioma[J]. Chin J Magn Reson Imaging, 2023, 14(12): 146-150, 171. DOI: 10.12015/issn.1674-8034.2023.12.026.
[13]
HWANG J, KANG J E, JEON S, et al. Transfer learning of deep neural networks pretrained using the ABCD dataset for general psychopathology prediction in Korean adolescents[J/OL]. Biol Psychiatry Cogn Neurosci Neuroimaging, 2025, 21: S2451-9022(25)00133-8. [2025-05-05]. https://linkinghub.elsevier.com/retrieve/pii/S2451-9022(25)00133-8. DOI: 10.1016/j.bpsc.2025.04.005.
[14]
BA Z C, ZHANG H X, LIU A Y, et al. Combination of DCE-MRI and NME-DWI via deep neural network for predicting breast cancer molecular subtypes[J/OL]. Clin Breast Cancer. 2024, 24(5): e417-e427 [2025-05-05]. https://linkinghub.elsevier.com/retrieve/pii/S1526-8209(24)00079-X. DOI: 10.1016/j.clbc.2024.03.006.
[15]
REN J, AN N, LIN C, et al. DeepPrep: an accelerated, scalable and robust pipeline for neuroimaging preprocessing empowered by deep learning[J]. Nat Methods, 2025, 22(3): 473-476. DOI: 10.1038/s41592-025-02599-1.
[16]
BAE S, CHOI Y S, AHN S S, et al. Radiomic MRI phenotyping of glioblastoma: improving survival prediction[J/OL]. Radiology, 2018 [2025-05-05]. https://pubs.rsna.org/doi/10.1148/radiol.2018180200?url_ver=Z39.88-2003&rfr_id=ori:rid:crossref.org&rfr_dat=cr_pub%20%200pubmed. DOI: 10.1148/radiol.2018180200.
[17]
JOHNSON D R, GIANNINI C, VAUBEL R A, et al. A radiologist's guide to the 2021 WHO central nervous system tumor classification: Part I-key concepts and the spectrum of diffuse gliomas[J/OL]. Radiology, 2023, 306(2): e229036 [2025-05-05]. https://pubs.rsna.org/doi/10.1148/radiol.213063?url_ver=Z39.88-2003&rfr_id=ori:rid:crossref.org&rfr_dat=cr_pub%20%200pubmed. DOI: 10.1148/radiol.229036.
[18]
REN J, ZHAI X, YIN H, et al. Multimodality MRI radiomics based on machine learning for identifying true tumor recurrence and treatment-related effects in patients with postoperative glioma[J]. Neurol Ther, 2023, 12: 1729-1743. DOI: 10.1007/s40120-023-00524-2.
[19]
LIU X, HAN T, WANG Y, et al. Prediction of O(6)-methylguanine-DNA methyltransferase promoter methylation status in IDH-wildtype glioblastoma using MRI histogram analysis[J/OL]. Neurosurg Rev, 2024, 47(1): 285 [2025-05-05]. https://link.springer.com/article/10.1007/s10143-024-02522-w. DOI: 10.1007/s10143-024-02522-w.
[20]
SAEED N, RIDZUAN M, ALASMAWI H, et al. MGMT promoter methylation status prediction using MRI scans? An extensive experimental evaluation of deep learning models[J/OL]. Med Image Anal. 2023, 90: 102989 [2025-05-05]. https://linkinghub.elsevier.com/retrieve/pii/S1361-8415(23)00249-9. DOI: 10.1016/j.media.2023.102989.
[21]
DONISELLI F M, PASCUZZO R, MAZZI F, et al. Quality assessment of the MRI-radiomics studies for MGMT promoter methylation prediction in glioma: a systematic review and meta-analysis[J]. Eur Radiol, 2024, 34(9): 5802-5815. DOI: 10.1007/s00330-024-10594-x.
[22]
LIANG Q, ZHANG H. Research progression of MRI radiomics in glioma[J]. Chin J Magn Reson Imaging, 2024, 15(2): 192-197. DOI: 10.12015/issn.1674-8034.2024.02.031.
[23]
LIN Y, CHAN S C W, CHUNG H Y, et al. A deep neural network for MRI spinal inflammation in axial spondyloarthritis[J]. Eur Spine J, 2024, 33(11): 4125-4134. DOI: 10.1007/s00586-023-08099-0.
[24]
ZHAO Y, TANG C, CUI B, et al. Automated segmentation of the human supraclavicular fat depot via deep neural network in water-fat separated magnetic resonance images[J]. Quant Imaging Med Surg, 2023, 13(7): 4699-4715. DOI: 10.21037/qims-22-304.
[25]
GAO T, LIU S, GAO E, et al. Automatic segmentation of laser-induced injury OCT images based on a deep neural network model[J/OL]. Int J Mol Sci, 2022, 23(19): 11079. [2025-05-05]. https://pmc.ncbi.nlm.nih.gov/articles/PMC9570418/. DOI: 10.3390/ijms231911079.
[26]
LOU Q, LUO X L, LIU H, et al. Predicting epidermal growth factor receptor gene mutation in lung adenocarcinoma based on CT traditional machine learning and deep neural network[J]. Chinese Journal of Medical Imaging, 2023, 31(9): 932-938. DOI: 10.3969/j.issn.1005-5185.2023.09.006.
[27]
ZHOU X, YUE X, XU Z, et al. PENet: Prior evidence deep neural network for bladder cancer staging[J]. Methods, 2022, 207: 20-28. DOI: 10.1016/j.ymeth.2022.08.010.
[28]
SHEN N, LV W, LI S, et al. Noninvasive Evaluation of the Notch Signaling Pathway via Radiomic Signatures Based on Multiparametric MRI in Association With Biological Functions of Patients With Glioma: A Multi-institutional Study[J]. J Magn Reson Imaging, 2023, 57(3): 884-896. DOI: 10.1002/jmri.28378.
[29]
MA W Q, CHEN K Y, YANG N, et al. Diagnostic value of machine learning based on multi-parameters of MRI radiomics to predict cervical lymph node status of papillary thyroid carcinoma[J]. Chin J Magn Reson Imaging, 2022, 13(10): 108-113. DOI: 10.12015/issn.1674-8034.2022.10.016.
[30]
SHI Z W, LIU Z Y. The challenges and solutions in radiomics study[J]. Chinese Journal of Radiology, 2022, 56(1): 9-11. DOI: 10.3760/cma.j.cn112149-20211111-00998.

PREV QSM and DKI for evaluation of iron deposition and microstructural alterations in gray matter nuclei of cerebral small vessel disease with mild cognitive impairment
NEXT Non-invasive prediction of HER-2 overexpression and low expression in NME-type breast cancer using multiparametric MRI radiomics combined with MRI features
  


LINK


Tel & Fax: +8610-67113815    E-mail: editor@cjmri.cn