Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2017 Oct;30(5):622-628.
doi: 10.1007/s10278-017-0009-z.

Residual Deep Convolutional Neural Network Predicts MGMT Methylation Status

Panagiotis Korfiatis  1 , Timothy L Kline  1 , Daniel H Lachance  2 , Ian F Parney  3 , Jan C Buckner  4 , Bradley J Erickson  5
Affiliations

Residual Deep Convolutional Neural Network Predicts MGMT Methylation Status

Panagiotis Korfiatis et al. J Digit Imaging. 2017 Oct.

Abstract

Predicting methylation of the O6-methylguanine methyltransferase (MGMT) gene status utilizing MRI imaging is of high importance since it is a predictor of response and prognosis in brain tumors. In this study, we compare three different residual deep neural network (ResNet) architectures to evaluate their ability in predicting MGMT methylation status without the need for a distinct tumor segmentation step. We found that the ResNet50 (50 layers) architecture was the best performing model, achieving an accuracy of 94.90% (+/- 3.92%) for the test set (classification of a slice as no tumor, methylated MGMT, or non-methylated). ResNet34 (34 layers) achieved 80.72% (+/- 13.61%) while ResNet18 (18 layers) accuracy was 76.75% (+/- 20.67%). ResNet50 performance was statistically significantly better than both ResNet18 and ResNet34 architectures (p < 0.001). We report a method that alleviates the need of extensive preprocessing and acts as a proof of concept that deep neural architectures can be used to predict molecular biomarkers from routine medical images.

Keywords: Deep learning; MGMT methylation; MRI.

PubMed Disclaimer

Figures

Fig. 1
Fig. 1
Visual depiction of the ResNet50 model. Conv, pool, and fc stand for convolutional, pooling, and fully connected layer, respectively. The pooling size used was 2 (denoted by “/2.” For instance, “256/2” means that 256 filters were used and the size of pooling layer was 2). So the first box (“7 × 7 conv, 64”) means that the convolutional kernel size was 7 × 7 and 64 filters. We then explicitly describe the following layer as a 2 × 2 pooling layer, but elsewhere in this figure we use the shorthand of “/2” in the box showing the layer. Solid lines (—) indicate identity and dashed lines (- - -) indicate cross-residual weighted connections
Fig. 2
Fig. 2
Input and activations (first 3 activation layers) of the selected ResNet50 network. Every box shows an activation map corresponding to the kernel function found by the network. Most values are near zero, but visible activations appear to reflect edges as well as important tissues
See this image and copyright information in PMC

References

    1. Johnson DR, O’Neill BP. Glioblastoma survival in the United States before and during the temozolomide era. J Neurooncol. 2011;107:359–364. doi: 10.1007/s11060-011-0749-4. - DOI - PubMed
    1. Ellingson BM, Wen PY, van den Bent MJ, Cloughesy TF. Pros and cons of current brain tumor imaging. Neuro Oncol. 2014;16(Suppl 7):vii2–vi11. doi: 10.1093/neuonc/nou224. - DOI - PMC - PubMed
    1. Weizman L, Ben-Sira L, Joskowicz L, Aizenstein O, Shofty B, Constantini S, Ben-Bashat D. Prediction of brain MR scans in longitudinal tumor follow-up studies. Med Image Comput Comput Assist Interv. 2012;15:179–187. - PubMed
    1. Law M, Young RJ, Babb JS, Peccerelli N, Chheang S, Gruber ML, Miller DC, Golfinos JG, Zagzag D, Johnson G. Gliomas: predicting time to progression or survival with cerebral blood volume measurements at dynamic susceptibility-weighted contrast-enhanced perfusion MR imaging. Radiology. 2008;247:490–498. doi: 10.1148/radiol.2472070898. - DOI - PMC - PubMed
    1. Jain R, Poisson LM, Gutman D, et al. Outcome prediction in patients with glioblastoma by using imaging, clinical, and genomic biomarkers: focus on the nonenhancing component of the tumor. Radiology. 2014;272:484–493. doi: 10.1148/radiol.14131691. - DOI - PMC - PubMed

MeSH terms