Fig. 1.

Overview of our classification framework. The MRI data were taken from ADNI data source and preprocessed by a combination of FSL and ANTs processing tools. The processed data flowed through a deep learning pipeline (top) and a XGBoost pipeline (bottom). Finally, the generated features of the two pipelines were concatenated with demographics features and cognitive scores to construct a combined features for the ensemble model (right). The ensemble model gave the final prediction (AD or CN).