Hi everyone,
I have a simple ( and yet confusing to me) question. Suppose we are only
interested in one-sample t-test for the second-level analysis to find the
patterns of activation within a group. The question is, what’s the
difference between the t-test function in Matlab (ttest.m) and the one-
sample t-test in SPM5. Note that my question concerns only calculating the
t-values for each voxel and I’m not concerned about choosing a threshold
yet.
In other words, if I do one-sample t-test on every voxel (assuming
different subjects are different observations) in Matlab, should it be the
same image as SPM_XT.img generated by SPM5?
Yet in another words, the standard ttest.m in Matlab, when used on each
voxel independently, will only consider the variance and mean of the same
exact voxel from all the subjects while computing the t-value, does SPM5
calculates the raw t-values using the same method or it incorporated the
correlation between voxels when doing so? Again, I want to remind you that
I am talking about only computing the raw t-values and I am not getting
into GRF or FDR correction discussion.
Ok Here’s the ultimate core point of question, what are the exact SPM5
formulas for computing the raw t-values associated with each Voxel (and
not correcting the threshold and etc.) in a one-sample t-test?
I hope I managed to get my point through.
Cheers
-------------------------------
Behrang Nosrat Makouei,
Masters of Applied Science,
School of Engineering Science,
MICA Lab (MIAL), TASC 9400, SFU.
cell: 778-885-5016
www.sfu.ca/~bnosratm
|