|
Abstract:
|
Alzheimer's disease affects the underlying structure and tissue of the brain. Magnetic Resonance Imaging enables researchers to observe tissue atrophy and abnormalities in the brain. These images may be broken up into over 200,000 tiny volumes, or voxels, of information. Larger studies are now focusing on serial images to detect disease-related patterns of brain changes. Typical analyses have reduced these data to a one-number summary, such as a volume, repeated over time. However, such a summary may not adequately capture the ways in which the brain changes. Alternative summaries may include decompositions of the serial voxel data that identify key features of change in size, location, and shape. We present a possible strategy for capturing patterns of change along with simulation results to illustrate the performance of such a strategy.
|
