Dear SPMers,
I want to compare one single patient's fMRI against a group of healthy controls. I understand from previous posts that a "Two-sample t-test" is basically ok for this and will deal correctly with such an extreme 'group size' of 1. I am also well aware that usually the contrast images are taken to the 2nd level analysis.
My question is: Would it be fundamentally wrong to use spmT maps rather than contrast images for this. Apparently, the difference between these two is, that the spmT maps have taken into account the residual variance from the ResMS image, resulting in lower T-values in areas with higher variance / error.
The reason I ask is, that this variance seems to be a major driving force for fMRI results in my data. I have seen several cases, where the contrast image does not reflect what I see in SPM 'Results' e.g. an asymmetric hippocampal activation, R>L when looking at 'Results', but symmetric hippocampal 'activation' in contrast image and a higher ResMS on the L, resulting in lower T.
Also the spmT maps look generally 'cleaner' i.e. have less (movement related) high intensity voxels at the outer edge of the dataset compared to contrast images - this also seems to be fixed by accounting for the high variance in these regions and lowering the T-value.
Why would I 'sacrifice' this additional information before taking it to the second level?
Or to put it the way: What would be a correct way to compare the smpT maps between groups?
Many thanks for some advice,
Christian
|