2.1. Data and Preprocessing

Data used in the preparation of this article were obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI, RRID:SCR_003007) database (adni.loni.usc.edu). The ADNI was launched in 2003 as a public-private partnership, led by Principal Investigator Michael W. Weiner, MD. The primary goal of ADNI has been to test whether serial magnetic resonance imaging (MRI), positron emission tomography (PET), other biological markers, and clinical and neuropsychological assessment can be combined to measure the progression of mild cognitive impairment (MCI) and early Alzheimer's disease (AD). For up-to-date information, see www.adni-info.org.

We included structural MRI data of all subjects with Alzheimer's disease (AD) and healthy controls (HCs) listed in the “MRI collection - Standardized 1.5T List - Annual 2 year". The subjects in the data set are labeled as AD if the Clinical Dementia Rating (CDR) score (Morris, 1993) was greater than 0.5. HCs are selected as those subjects with a CDR score of 0. In total, we included 969 individual scans (475 AD, 494 HC) of 193 AD patients and 151 HCs (up to three time points). All scans were acquired with 1.5 T scanners at various sites and had undergone gradient non-linearity, intensity inhomogeneity and phantom-based distortion correction. We downloaded T1-weighted MPRAGE scans and non-linearly registered them to the 1mm resolution 2009c version of the ICBM152 reference brain using Advanced Normalization Tools (ANTs²). This has been done to (1) ensure a relative alignment across subjects, (2) allow the convolutional neural network to extract more robust features, and (3) be able to analyze the heatmaps in a common space. For the region-wise analysis of heatmaps, we used the Scalable Brain Atlas by Neuromorphometrics Inc. (Bakker et al., 2015) available in SPM12³. A list of all areas included can be found in the SPM12 package.