Residual convolutional neural network for the determination of IDH status in low- and high-grade gliomas from mr imaging

Ken Chang, Harrison X. Bai, Hao Zhou, Chang Su, Wenya Linda Bi, Ena Agbodza, Vasileios K. Kavouridis, Joeky T. Senders, Alessandro Boaro, Andrew Beers, Biqi Zhang, Alexandra Capellini, Weihua Liao, Qin Shen, Xuejun Li, Bo Xiao, Jane Cryan, Shakti Ramkissoon, Lori Ramkissoon, Keith LigonPatrick Y. Wen, Ranjit S. Bindra, John Woo, Omar Arnaout, Elizabeth R. Gerstner, Paul J. Zhang, Bruce R. Rosen, Li Yang, Raymond Y. Huang, Jayashree Kalpathy-Cramer

Research output: Contribution to journalArticle

84 Scopus citations

Abstract

Purpose: Isocitrate dehydrogenase (IDH) mutations in glioma patients confer longer survival and may guide treatment decision making. We aimed to predict the IDH status of gliomas from MR imaging by applying a residual convolutional neural network to preoperative radiographic data. Experimental Design: Preoperative imaging was acquired for 201 patients from the Hospital of University of Pennsylvania (HUP), 157 patients from Brigham and Women's Hospital (BWH), and 138 patients from The Cancer Imaging Archive (TCIA) and divided into training, validation, and testing sets. We trained a residual convolutional neural network for each MR sequence (FLAIR, T2, T1 precontrast, and T1 postcontrast) and built a predictive model from the outputs. To increase the size of the training set and prevent overfitting, we augmented the training set images by introducing random rotations, translations, flips, shearing, and zooming. Results: With our neural network model, we achieved IDH prediction accuracies of 82.8% (AUC ¼ 0.90), 83.0% (AUC ¼ 0.93), and 85.7% (AUC ¼ 0.94) within training, validation, and testing sets, respectively. When age at diagnosis was incorporated into the model, the training, validation, and testing accuracies increased to 87.3% (AUC ¼ 0.93), 87.6% (AUC ¼ 0.95), and 89.1% (AUC ¼ 0.95), respectively. Conclusions: We developed a deep learning technique to noninvasively predict IDH genotype in grade II–IV glioma using conventional MR imaging using a multi-institutional data set.

Original languageEnglish (US)
Pages (from-to)1073-1081
Number of pages9
JournalClinical Cancer Research
Volume24
Issue number5
DOIs
StatePublished - Mar 1 2018

ASJC Scopus subject areas

  • Oncology
  • Cancer Research

Fingerprint Dive into the research topics of 'Residual convolutional neural network for the determination of IDH status in low- and high-grade gliomas from mr imaging'. Together they form a unique fingerprint.

  • Cite this

    Chang, K., Bai, H. X., Zhou, H., Su, C., Bi, W. L., Agbodza, E., Kavouridis, V. K., Senders, J. T., Boaro, A., Beers, A., Zhang, B., Capellini, A., Liao, W., Shen, Q., Li, X., Xiao, B., Cryan, J., Ramkissoon, S., Ramkissoon, L., ... Kalpathy-Cramer, J. (2018). Residual convolutional neural network for the determination of IDH status in low- and high-grade gliomas from mr imaging. Clinical Cancer Research, 24(5), 1073-1081. https://doi.org/10.1158/1078-0432.CCR-17-2236