Spatial-Temporal Models for Image Data Analyses

• Activity Number: 362
• Type: Contributed
• Date/Time: Tuesday, August 6, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Imaging
• Author(s): Chun-Jung Huang*+ and Laurel Beckett and Danielle Harvey
• Companies: UC Davis and UC Davis and UC Davis
• Keywords: Spatial-temporal ; imaging data ; Alzheimer ; Flobetapir ; amyloid-beta ; voxel

Abstract

Alzheimer's disease is the most common form of dementia for people over 65. Amyloid-beta plaques and neurofibrillary tangles are the pathologic hallmarks of Alzheimer's disease in brain tissue seen in autopsy. Through chemical agents, such as Florbetapir F18 (AV-45), Positron Emission Tomography can quantify amyloid-beta in the human brain. Our goal is to develop statistical models that characterize the amyloid-beta patterns in brains cross-sectionally and longitudinally by using the information in brain images at the voxel-level, while accounting for the spatial and temporal correlation. The biggest challenge falls in the complexity of the interaction between spatial and temporal correlations. Also, anatomic structure and biological pathways in the brain may induce correlations between voxels. Different models are proposed to characterize the spatial pattern and temporal trajectory. Simulation results will be presented to illustrate the small-sample (and small region) properties and to compare performance across models.