The ArtRepair toolbox makes some approximations to handle noisy
fMRI data from high motion subjects. One is that rapid motions
or transient noise may cause some volumes to have bad data values that
fall far outside the noise assumptions used in the GLM. The
program interpolates through the bad data to provide a nicer
time series for the GLM. Second is that the interpolated volumes
should be counted less in the estimation process by deweighting
those volumes by a factor of 100 (relative to the other volumes).
However, there is a side effect. SPM normally tunes the weights
for each volume (as described in Guillaume Flandin's response),
and specifying an apriori deweighting (SPM.xW.W) precludes
that optimization (according to the SPM documentation). Consequently,
deweighting unfortunately prevents an SPM optimization from occurring.
That is clearly a loss, but the assumption is that repair and deweighting
may still be better for strongly non-Gaussian data. (This is only
an assumption, and there is no proof.)
For group studies, only the single subject estimates are passed up to
the group level. In this case, the repair is the helpful feature, and
there seems to be little difference between estimates generated
with or without deweighting.
Generally, artifact repair and deweighting works well for blips of high motion
or other short duration artifacts. Motion regressors are more effective
when there are many scans with motion. Most importantly, review the single
subject contrast maps to quality check that the estimates are reasonable,
because high motion fMRI data sets are a challenge to analyze by any algorithm.
Best regards,
Paul
----- Original Message -----
From: "Elizabeth Liddle" <[log in to unmask]>
To: [log in to unmask]
Sent: Tuesday, May 5, 2009 4:53:39 AM GMT -08:00 US/Canada Pacific
Subject: [SPM] SPM.xX.W
Can someone explain what these weights are, and what they are used for?
The reason I ask is that I have been using ArtRepair, and I notice that
they are replaced by a much simpler matrix, and I wonder what we are
losing, if anything.
Thanks
Elizabeth Liddle
This message has been checked for viruses but the contents of an attachment
may still contain software viruses, which could damage your computer system:
you are advised to perform your own checks. Email communications with the
University of Nottingham may be monitored as permitted by UK legislation.
--
Paul K. Mazaika, PhD.
Center for Interdisciplinary Brain Sciences Research
Stanford University School of Medicine
Office: (650)724-6646 Cell: (650)799-8319
CONFIDENTIALITY NOTICE: Information contained in this message and any
attachments is intended only for the addressee(s). If you believe
that you have received this message in error, please notify the
sender immediately by return electronic mail, and please delete it
without further review, disclosure, or copying.
|