If you want to do amygdala masks in native space, the best approach is to use an appropriate method for hand-drawing an amygdala mask for each. This, however, requires experience to make sure the masks are accurate.
Alternately, you can use a number of brain segmentation algorithms. Here is a random example I found by googleoling automated amygdala segmentation:
There may be several others. They may have varying levels of availability and usability. This is the approach I would generally recommend (find an auto segmentation method).
The easiest, and least reliable, method it to take an amygdala mask in MNI space, and apply a reverse transform or warp to move it into that participant’s native space. This involves calculating the transform from native to MNI space for that individual, and then applying the inverse of that transform to a mask file in MNI space, using the –nearest-neighbor flag in FNIRT.
I’d post an example command but I am not at a computer with FSL. Check FNIRT’s documentation on inverse transforms. And do some googling and reading papers where people do subject-specific amygdala masks to get a sense of what has worked.
Good Luck!
Colin Hawco, PhD
Neuranalysis Consulting
Neuroimaging analysis and consultation
Dear all, To create a pre-thresholding mask, say if I have 10 subjects, how can I create a pre-thresholding mask of amygdala in the native fMRI space? Is the idea to have an average subject first, and create the amygdala mask in that subject? Best, Daniel
