Arterial Spin Labelling (ASL) is an MRI method to measure cerebral blood flow with potential to assist early dementia diagnosis. Here, ASL data acquired from patients with young onset Alzheimer’s disease (AD) was analysed, using both a novel region based statistical approach and voxel based machine learning. This is the first study to analyse ASL data from patients with Posterior Cortical Atrophy using machine learning. Both approaches are shown to identify regions known to be affected by AD. Inter-region analysis suggests the parietal lobe is the most useful benchmark region, to separate region specific hypoperfusion from global perfusion changes.