This paper describes a method for processing functional magnetic reson
ance images that suppresses signal changes originating from macroscopi
c reins visible in acquired magnetic resonance angiograms. Finger tapp
ing experiments were performed on a 1.5-T scanner and the response was
evaluated with voxel-by-voxel cross-correlation of the time course wi
th a sinusoid at the paradigm frequency. After applying a vascular mas
h to suppress signal changes under macroscopic vessels, the vascular a
nd non vascular subpopulations of the data were compared. By visual in
spection, the method was found to remove extracortical activation whil
e preserving activation in the parenchyma. The observed higher signal
amplitudes and temporal phase lags of the vascular population agree wi
th theoretical models and previous studies. A significant portion of n
egatively correlated voxels occurs adjacent to through-plane vessels.
Finally, comparing the centers of mass of the activated area before an
d after vascular suppression showed significant shifts in some subject
s. (C) 1998 Academic Press.