Identifying the regional substrates predictive of Alzheimer’s disease progression through a convolutional neural network model and occlusion

Abstract

AbstractProgressive brain atrophy is a key neuropathological hallmark of Alzheimer’s disease (AD). However, atrophy patterns along the progression of AD are diffuse and variable. Consequently, identifying the major regional atrophy patterns underlying AD progression is challenging. In the current study, we propose a method that evaluates the degree to which specific regional atrophy are predictive of AD progression, while holding all other atrophy changes constant. We first trained a dense convolutional neural network model to differentiate individuals with mild cognitive impairment (MCI) who progress to AD vs. those with a stable MCI diagnosis. Then, we retested the model multiple times, each time occluding major regions from the model’s testing set’s input. This revealed that the hippocampus, fusiform, and inferior temporal gyri, were the strongest predictors of AD progression, in agreement with established staging models. These results shed light on the major regional patterns of atrophy predictive of AD progression.