A Practical Alzheimer Disease Classifier via Brain Imaging-Based Deep Learning on 85,721 Samples

Beyond detecting brain lesions or tumors, comparatively little success has been attained in identifying brain disorders such as Alzheimers disease (AD), based on magnetic resonance imaging (MRI). Many machine learning algorithms to detect AD have been trained using limited training data, meaning they often generalize poorly when applied to scans from previously unseen populations. Here we aimed to build a practical brain imaging-based AD diagnostic classifier using deep learning/transfer learning on dataset of unprecedented size and diversity. We pooled MRI data from more than 217 sites/scanners to constitute the largest brain MRI sample to date (85,721 scans from 50,876 participants). Next, we applied a state-of-the-art deep convolutional neural network, Inception-ResNet-V2, to build a sex classifier with high generalization capability. The sex classifier achieved 94.9% accuracy and served as a base model in transfer learning for the objective diagnosis of AD. After transfer learning, the model fine-tuned for AD classification achieved 91.3% accuracy in leave-sites-out cross-validation on the Alzheimer9s Disease Neuroimaging Initiative (ADNI) dataset and 94.2%/87.9% accuracy for direct tests on two unseen independent datasets (AIBL/OASIS). When this AD classifier was tested on brain images from unseen mild cognitive impairment (MCI) patients, MCI patients who finally converted to AD were 3 times more likely to be predicted as AD than MCI patients who did not convert (65.2% vs 20.6%). Predicted scores from the AD classifier showed significant correlations with illness severity. In sum, the proposed AD classifier could offer a medical-grade biomarker that could be integrated into AD diagnostic practice. Our trained model, code and preprocessed data are freely available to the research community.