“…Traditionally, classification frameworks focus only on the brain regions that are commonly affected by the disease instead of searching for information patterns throughout the whole brain. This reduces the number of features to be processed, which partially solves the curse of dimensionality present in most statistical classifications [16]. Once the brain region is delimited, the easiest ap-proach is to use the intensity of the voxels as predictors of the different classes to be modelled, using them as inputs of the classifier.…”