Automatic Prediction of MGMT Status in Glioblastoma via Deep Learning-Based MR Image Analysis

• Published in: BioMed research international 2020, Vol.2020 (2020), p.1-9
• Main Authors: Yang, Ruimeng, Liu, Zaiyi, Wei, Xinhua, Jiang, Xinqing, Cheng, Zixuan, Zhang, Zhongping, Li, Haixia, Fu, Yan, Zhang, Tianjing, Tong, Yichen, Zeng, Min, Chen, Xin, Xu, Xiangdong
• Format: Article
• Language: English
• Subjects: Analysis; Annotations; Biomarkers; Biomedical research; Brain - diagnostic imaging; Brain - pathology; Brain cancer; Brain Neoplasms - diagnostic imaging; Brain Neoplasms - genetics; Brain Neoplasms - pathology; Care and treatment; Classification; Datasets; Deep Learning; Diagnosis; DNA methylation; DNA Methylation - genetics; DNA Modification Methylases - genetics; DNA Repair Enzymes - genetics; Glioblastoma; Glioblastoma - diagnostic imaging; Glioblastoma - genetics; Glioblastoma - pathology; Glioblastoma multiforme; Glioma; Health aspects; Humans; Image analysis; Image classification; Image contrast; Image enhancement; Image Interpretation, Computer-Assisted - methods; Image processing; Image segmentation; Imaging Genomics - methods; Machine learning; Magnetic resonance imaging; Magnetic Resonance Imaging - methods; Medical diagnosis; Medical imaging; Methods; Methylation; Methylguanine; Methyltransferases; O6-methylguanine-DNA methyltransferase; Patients; Pipeline design; Predictions; Promoter Regions, Genetic - genetics; Risk factors; ROC Curve; Tumor Suppressor Proteins - genetics; Tumors
• ISSN: 2314-6133; 2314-6141
• DOI: 10.1155/2020/9258649

Methylation of the O6-methylguanine methyltransferase (MGMT) gene promoter is correlated with the effectiveness of the current standard of care in glioblastoma patients. In this study, a deep learning pipeline is designed for automatic prediction of MGMT status in 87 glioblastoma patients with contrast-enhanced T1W images and 66 with fluid-attenuated inversion recovery(FLAIR) images. The end-to-end pipeline completes both tumor segmentation and status classification. The better tumor segmentation performance comes from FLAIR images (Dice score, 0.897±0.007) compared to contrast-enhanced T1WI (Dice score, 0.828±0.108), and the better status prediction is also from the FLAIR images (accuracy, 0.827±0.056; recall, 0.852±0.080; precision, 0.821±0.022; and F1 score, 0.836±0.072). This proposed pipeline not only saves the time in tumor annotation and avoids interrater variability in glioma segmentation but also achieves good prediction of MGMT methylation status. It would help find molecular biomarkers from routine medical images and further facilitate treatment planning.