Dear Simone and list,
Unfortunately, the two different head coils were used to acquire two
non-overlapping data sets (ie condition A patients and controls in one coil
and condition B patients and controls with another coil).
Would it be reasonable to compare the two groups of patients to each other
and deal with the head coil confound by including as a regressor of no
interest a whole brain volume (ie using the BPM toolbox)? That volume could
be each subject's SNR map derived from the same task, or their main effect
activation map for the task. This way, it will effectively model possible
confounds of SNR differences separately at each voxel for each subject....
Amit
|