The proposed project will apply advanced computer analysis methods to magnetic resonance images (MRI) of mildly cognitively impaired (MCI) patients. MCI is believed to often be an early Alzheimer’s Disease (AD) stage for many patients, since MCI patients progress to AD at a rate of approximately 15% per year, albeit many remain stable for many years. Our main goals are to Aim 1: To measure how the amount of brain tissue in various brain regions diminishes over time, and to precisely quantify the differences between MCI patients that convert to AD within a 3-year period from those that remain stable. Aim 2: To apply pattern recognition methods to the measurements of Aim 1, and determine which aspects of brain change over time allow us to predict whether an MCI patient will progress to AD, or not. Our main hypotheses are that 1) The patterns of brain atrophy are significantly different between MCI patients that convert to AD and those that don’t. 2) The application of pattern recognition methods to measurements of brain atrophy will allow us to predict with high accuracy which patients are likely to progress from MCI to AD, and which are likely to remain stable. Such predictions are critical for the management of patients with cognitive decline, since early diagnosis of AD significantly increases the likelihood that potential treatments will work, but also because many potential treatments currently have serious side-effects, thereby making it necessary for a physician to know whether a particular patient needs to undergo the treatment, or he/she is likely to remain stable without treatment. We will utilize MRI data from a large NIH-funded study, the Alzheimer’s Neuroimaging Initiative, therefore no funds for patient recruitment and image acquisition are needed.