Hello,
I'm trying to be careful about the interpretability of voxel values in
the slice (z) direction following smoothing (because I'm processing some
ASL data with only 24 slices). As you all know, following Realignment,
it is common for a certain number of voxels in the top and/or bottom
slices to be masked to 0, due to movement.
The Smooth module has an option to respect that masking (by setting
Implicit Masking = Yes), so that voxels that were 0 following
realignment are still 0 after smoothing. That certainly is a useful
option.
Further the description of spm_smooth says:
"Boundary conditions assume data does not exist outside the image in z
(i.e. the kernel is truncated in z at the boundaries of the image
space)".
However, what happens if there are 0's at the extreme slices in the z
direction prior to smoothing? Are those 0's used to define an EMPIRICAL
boundary for the smoothing kernel? If not, it seems to me that the 0's
that are present following Realignment will artificially lower the
values of the adjacent non-zero voxels, thereby reducing the
interpretability and validity of those non-zero voxels themselves.
Empirically, it appears that 0's within the image space ARE being
treated as legitimate values as far as the smoothing operation is
concerned, even if Implicit Masking = Yes (i.e,. the kernel is not
truncated/adjusted for underlying 0 voxels).
Am I correct about this? If so, is there any way to avoid this problem
(other than not smoothing at all)?
thanks,
-Mike H.
--
Michael Harms, Ph.D.
--------------------------------------------------------------------
Conte Center for the Neuroscience of Mental Disorders
Washington University School of Medicine
Department of Psychiatry, Box 8134
Renard Hospital, Room 6604 Tel: 314-747-6173
660 South Euclid Ave. Fax: 314-747-2182
St. Louis, MO 63110 Email: [log in to unmask]
--------------------------------------------------------------------
|