Alzheimer’s disease (AD) - the most common form of dementia, is a term used for accelerated memory loss and cognitive abilities enough to severely hamper day-to-day activities. One of the most globally accepted markers for AD is atrophy, in mainly the brain parenchyma. The goal of the PhD project is to develop a deformation-based atrophy estimation pipeline that is reliable and stable. We investigate the application of image registration using freeform deformation and stationary velocity fields in atrophy estimation. In the process, we propose a new multi-scale model for the transformation model and a new way to estimate atrophy from a deformation field. We demonstrate the performance of the proposed solution but applying it on the publicly available Alzheimer’s disease neuroimaging data (ADNI) initiative and compare to existing state-of-art atrophy estimation methods.