In this project, structural MRI data will be obtained at baseline from controls and non-demented subjects with cognitive impairments who will be recruited through the Clinical Core and followed over time. Subjects will be evaluated longitudinally with MRI when they cross specific transition boundaries, as defined by the Clinical Core. The overall goal of the MRI project is to determine whether structural MRI data can be used to predict which non-demented individuals with cognitive problems will develop progressive cognitive decline, which will progress to the point where they meet criteria for AD, and which will remain stable. In addition, we will seek to determine the evolution of brain changes that characterize prodromal AD, and how these measures relate to the clinical characteristics of the subjects. Underlying these goals is the assumption that structural MRI data can serve as an indirect measure of the neuronal loss that occurs when neuritic plaques and neurofibrillary tangles accumulate during the development of Alzheimer's disease (AD). In the proposed funding cycle we will use manually drawn regions of interest (R01), as we have done in the past, to evaluate atrophy in selected brain regions. We are also proposing to enhance and apply a new automated method for identifying ROIs automatically, and to compare this method to the manual measurements. Both types of MRI data will e used by other projects in the Program Project, and thus an additional responsibility of this Project is to distribute MRI images, and regions of interest derived from them, to investigators in Project by Johnson (SPECT) and Project by Sperling (fMRI). In addition, we will seek to determine the relationship between SPECT measures and other information obtained on the same individuals by the cores and the projects, including clinical data, genotype status, volumetric MRI, and fMRI signal change.