Dear SPM Users,
I am facing a
trouble
related to motion correction, and possibly, the processing stage.
In the past I
used the
procedures suggested in the SPM5 manual with the use of the parameter
file,
generated during the segmentation step, for normalizing the functional
images.
It worked fairly well.
But now I am
reanalyzing the
same data using slightly different preprocessing procedures. The new
preprocessing steps are as follows:
1. SLICE TIMING -
2
sessions; # of slices: 30; TR:2; TA: 1.933; ascending interleaved;
reference:
15.
2. REALIGNMENT
ESTIMATE AND
RESLICE - 2 sessions; register to first.
3. COREGISTER -
meanaf.img +
subject's own T1 3D structural image.
4. NORMALIZE
ESTIMATE AND
RESLICE - 1 subject; source: original T1 3D img; images to write: all
realigned
img + T1 3D img; voxel size: resampled from
3x3x5 to 3x3x3;
template:
MNI T1.
5. SMOOTH - all
normalized
images.
HERE IS THE
PROBLEM
This partly new
preprocessing procedure seems to work well through all the steps.
However,
after performing the first-level analysis and estimation and generation
of
contrast images, when I check the results, there seems to be some
activation
displacement, I mean, some activations apparently are located off the
brain
tissue.
I have tried
other
normalization methods as well, such as the one proposed on Christoff's
lab
website. But the same result awfully appears.
My guess is that
the problem
arises from the coregistration or normalization.
I don't know
whether
slice/voxel size resampling would cause such problem.
I have attached a
PDF file
with some images from the preprocessing stages and the results.
Would any fellow
comment on
the things I have done so far?
I would like to
know how
good are these steps above and which should be the best way?
I would very much
appreciate
your comments.
Alan Fermin, PhD
student
Okinawa Institute
of Science and Technology
Neural
Computation Unit
http://www.nc.irp.oist.jp/
[log in to unmask]
Okinawa, Japan