A multivariate approach for mapping longitudinal changes in amyloid deposition

Introduction: The temporal nature of amyloid deposition in Alzheimer's disease (AD) is of much research interest. PET imaging has shown evidence of amyloid deposition in cognitively normal (CN) subjects and numerous longitudinal PET amyloid imaging studies are ongoing. Partial least squares (PLS) is a multivariate analysis technique that can be used to explore regional patterns in functional brain imaging data that best distinguish different states [1,2] and we have previously used PLS to explore relationships among regional baseline PiB measures [3]. This work applies PLS at the voxel level to explore its utility to map patterns of longitudinal change in amyloid deposition.