| |
Abstract:
Abstract: Atrophy is a prominent feature of aging. Fast and
reliable characterization of atrophy may provide clinically useful
information, predict certain components of cognitive impairment,
and help constrain functional measurements based on hemodynamic
methods. We report here a fully automated method for characterizing
atrophy. 85 total subjects from 5 separate groups (young <
35 years old, middle aged 40-60, nondemented old 65+, and demented
old 65+) were imaged using multiple acquisitions of a high
resolution T1-weighted sequence (MP-RAGE). Images were (1)
corrected for inter-scan movement, (2) warped into a standard atlas
space based on the Talairach and Tournoux atlas (1988), (3)
corrected for intensity inhomogeneity, (4) masked to include only
the brain, and (5) segmented based on fuzzy cluster analysis into
gray, white and CSF. The complete, unsupervised analysis procedure
took less than 10 minutes. Two metrics proved useful in quantifying
atrophy: the relative difference between the signal intensities of
the gray and white matter peaks and the relative percentage of CSF
and white matter voxels. Results indicated that atrophy occurs in
healthy aging and more so in dementia of the Alzheimer type (DAT).
We discuss these findings in the context of cognitive assessment
and functional brain imaging methods.
|
