In conducting a regression analysis we have run into a strange problem.
After doing a simple regression between a personality score and
activation in our condition of interest, the results show a peak voxel
at (-30, 63, 18). We can extract these values from the voxel for
individual subjects by querying Matlab for the values of y, and the
scatterplot of the values against personality scores is well-behaved.
However, when I attempt to extract values from this same voxel under
other conditions by using a script similar to the one posted by John
Ashburner, SPM responds with a series of "NaN"s, as if the voxel was
outside of the brain.
I then attempted to extract these values from the first condition of
interest that yielded the peak voxel by using the script, and it again
tells me that the values are NaN's, which does not make sense
considering this is the condition which gave the voxel in question as
its most significant result.
When I examine the contrast images, the voxel in question is outside of
the brain; however, in the probability (T) images, the voxel is within
the brain.
Can anyone shed some light on what may be going wrong here? We are
running SPM2 on RedHat 9, Matlab 6.5.1 and have all the most recent
updates.
Thank you very much,
Deidre
--------------------------------------------
Deidre Kolarick
Department of Psychology
Yale University
Box 208205
New Haven, CT 06520-8205
[log in to unmask]
|