Dear Jerry and Michiru:
There is a way to mask an individual time-series in SPM, however, doing so is not necessarily a good thing and this depends on the mask you want to use. As Michiru notes he wants to mask
the brain to get rid of extraneous voxels so the number of multiple comparisons falls. This way the remaining areas, presumably "of interest," would more likely pass various significance
thresholds. The problem with this idea is that random field calculations are based not just on the volume of the field, and smoothness of the field, but on the shape of the field. Thus
if one was to calculate the expected euler characteristic of a sphere vs. a thin ribbon of equal volume (e.g. the gray matter ribbon) one would find that it was much higher, and the
significance levels much lower, for the thin ribbon. This is because the euler characteristic calculation includes 1-D, 2-D and 3-D components for the random field. Equivalently, while it
might seem helpful to reduce the comparisons in a brain by restricting analyses to the gray matter, it turns out that significance levels drop when one only uses gray matter. Now this
isn't true in the absolute sense, in that if you have a "small enough" region of gray matter that has been smoothed, then it might make sense to mask out everything else. Conversely at
that point one could use small volume correction instead. The other way to do this is to mask out some large region of the brain such as the cerebellum if you don't want to look there at
all.
Some more info about this is available here:
http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind0006&L=spm&D=0&P=2499
How to do this masking for a single subject.
make your mask based on a traced region of interest, or particular slices, or segmented parts of brain, etc. MRIcro is good software (and free- thanks to chris rorden) for making
ROI's. see: http://www.psychology.nottingham.ac.uk/staff/cr1/mricro.html
Take the first image of your time series and mask it with your mask by using imcalc in SPM.You would select your mask and the first volume from your time series (make sure they have
the same dimensions). Assume i1=fmri, and i2=mask. Then if your mask has ones for areas you want to include and zeros elsewhere use an equation such as i1(i2).
Set up your analysis and include the masked functional volume as the first volume in the time series. SPM only includes voxels in the analysis that have signal in all images so during
the analysis SPM willl now "mask" the rest of the time series for you.
voila.
Darren
--
----------------------------------------------------------------------
Darren R. Gitelman, M.D. E-mail: [log in to unmask]
Cognitive Neurology and Voice: (312) 908-9023
the Alzheimer¹s Disease Center Fax: (312) 908-8789
http://www.brain.nwu.edu
Northwestern Univ., 320 E. Superior St., Searle 11-470, Chicago, IL 60611
----------------------------------------------------------------------
|