Funny you should ask -- search the archives for "bug featquery" and
you'll find a thread on this topic from just last week.
The answer is that the mean, median, and percentile statistics from
featquery do NOT exclude zeros within the statistic image that fall
under the specified mask. We encountered the exact same situation as
you -- namely a mask defined in standard space, but zeros under that
mask in the patient (native) space. If you have enough such zeros, the
results of featquery, as currently written, are basically meaningless
(with one exception, which is the number of voxels output).
The solution is to either:
1) Create your own modified version of featquery (and its GUI) that uses
the appropriate "capitalized" flags in the call to 'fslstats', in order
to exclude the zeros
2) Get the mask into native space without featquery and then calculate
the stats of interest directly, again using the appropriate capitalized
flags in 'fslstats'
3) Run featquery as is just to get the mask in native space, but then
recalculate the stats of interest as in (2).
On Mon, 2010-06-21 at 21:05 +0100, Gabor Perlaki wrote:
> Hi all,
> I have a mask (for instance 'prefrontal cortex') in standard space, but I have no measurements at some parts of that mask in my patient space. How featquery handles it? It would be good if featquery neglected those parts of my mask, where there were no measurements in my patients space, and calculated the 'mean cope', 'mean z-stat' etc. without those parts.
> Does anybody really know how is it in featquery?
> Thanks, Gabor