Thanks Geoff. But how to handle the motion between the labeling and
control images? I do find motion between the initial label/control
pairs.
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]]
On Behalf Of Geoffrey K Aguirre
Sent: Wednesday, July 13, 2005 7:51 AM
To: [log in to unmask]
Subject: Re: [SPM] Perfusion MRI processing
On Jul 13, 2005, at 1:03 PM, Ze Wang wrote:
> For ASL images, I always realign them before subtraction, and
> for realignment, I just follow the general procedure, although
> Geoffrey
> Aguirre recommends the separate realignment.
Because the label and control images have systematically
different image intensities, it seems wise to realign them back to
separate label and control initial images. I have, however, never
tested to see if this makes any substantive difference.
> According to
> JJ Wang's paper, spatial smoothing is very important for ASL images.
The concept here is that whereas spatial smoothing of BOLD data
is associated with some deleterious effects upon the intrinsic
temporal noise structure, this unwanted effect is absent for ASL
data. In other words, spatial smoothing of ASL data provides the full
benefit that would be expected. There is no necessary incentive to
smooth ASL data, however, beyond what would be desired to maximize
detection of spatially extended signal changes.
For those interested in the technical details: BOLD fMRI data
have enhanced noise at low temporal frequencies. These low-frequency
temporal fluctuations tend to share phase across space, so spatial
smoothing of BOLD data acts to enhance the noise present at low
temporal frequencies.
Geoff
--
Geoffrey Karl Aguirre, M.D., Ph.D.
[log in to unmask]
Assistant Professor of Neurology
Hospital of the University of Pennsylvania
Center for Cognitive Neuroscience
Philadelphia, PA
|