PT - JOURNAL ARTICLE AU - Nikhil J. Dhinagar AU - Sophia I. Thomopoulos AU - Conor Owens-Walton AU - Dimitris Stripelis AU - Jose Luis Ambite AU - Greg Ver Steeg AU - Daniel Weintraub AU - Philip Cook AU - Corey McMillan AU - Paul M. Thompson TI - 3D Convolutional Neural Networks for Classification of Alzheimer’s and Parkinson’s Disease with T1-Weighted Brain MRI AID - 10.1101/2021.07.26.453903 DP - 2021 Jan 01 TA - bioRxiv PG - 2021.07.26.453903 4099 - http://biorxiv.org/content/early/2021/10/05/2021.07.26.453903.short 4100 - http://biorxiv.org/content/early/2021/10/05/2021.07.26.453903.full AB - Parkinson’s disease (PD) and Alzheimer’s disease (AD) are progressive neurodegenerative disorders that affect millions of people worldwide. In this work, we propose a deep learning approach to classify these diseases based on 3D T1-weighted brain MRI. We analyzed several datasets including the Parkinson’s Progression Markers Initiative (PPMI), an independent dataset from the University of Pennsylvania School of Medicine (UPenn), the Alzheimer’s Disease Neuroimaging Initiative (ADNI), and the Open Access Series of Imaging Studies (OASIS) dataset. The UPenn and OASIS datasets were used as independent test sets to evaluate the model performance during inference. We also implemented a random forest classifier as a baseline model by extracting key radiomics features from the same T1-weighted MRI scans. The proposed 3D convolutional neural network (CNN) model was trained from scratch for the classification tasks. For AD classification, the 3D CNN model achieved an ROC-AUC of 0.878 on the ADNI test set and an average ROC-AUC of 0.789 on the OASIS dataset. For PD classification, the proposed 3D CNN model achieved an ROC-AUC of 0.667 on the PPMI test set and an average ROC-AUC of 0.743 on the UPenn dataset. Model performance was largely maintained when using only 25% of the training dataset. The 3D CNN outperformed the random forest classifier for both the PD and AD tasks. The 3D CNN also generalized better on unseen MRI data from different imaging centers. These approaches show promise for screening of PD and AD patients using only T1-weighted brain MRI, which is relatively widely available. This model with additional validation could also be used to help differentiate between challenging cases of AD and PD when they present with similarly subtle motor and non-motor symptoms.Competing Interest StatementDr. McMillan receives research funding outside the submitted work from Biogen, Inc. and provides consulting services for personal fees from Invicro and Axon Advisors on behalf of Translational Bioinformatics, LLC. He also receives an honorarium as Associate Editor of NeuroImage: Clinical. Dr. Thompson receives research funding outside the submitted work from Biogen, Inc. and has received consulting fees from Kairos Venture Capital. Dr. Weintraub has received research funding or support from Michael J. Fox Foundation for Parkinson's Research, Alzheimer's Therapeutic Research Initiative (ATRI), Alzheimer's Disease Cooperative Study (ADCS), the International Parkinson and Movement Disorder Society (IPMDS) the National Institute on Aging (NIA) and the U.S. Department of Veterans Affairs; honoraria for consultancy from Acadia, Aptinyx, CHDI Foundation, Clintrex LLC (Otsuka), Eisai, Great Lake Neurotechnologies, Janssen, Sage, Scion, Signant Health, Sunovion and Vanda; and license fee payments from the University of Pennsylvania for the QUIP and QUIP-RS.