Sudipto Dolui1,2, David A. Wolk2, and John A. Detre1,2
1Department of Radiology, University of Pennsylvania, Philadelphia, PA, United States, 2Department of Neurology, University of Pennsylvania, Philadelphia, PA, United States
We
propose SCRUB, a data cleaning technique to improve cerebral blood flow (CBF)
estimation based on arterial spin labeling (ASL) data. The method consists of
(i) an outlier detection and removal stage and (ii) a subsequent voxel-wise
empirical robust Bayesian estimation step. Compared to alternative options,
SCRUB provided (i) better retest agreement between CBF values obtained from ASL
scans of elderly Controls in ADNI database acquired 3 months apart, and (ii)
better discrimination between Controls and patients with Alzheimer’s disease
(AD) based on CBF values in several regions of interest which are sensitive to
AD related changes.
