Hi,

We are beginning to collect images from multiple sequence types in each subject, BOLD epi, T1-weighted anatomicals, diffusion weighted/tensor images, and spectroscopic images (soon?) from several groups of subjects, like bipolar and healthy subjects.

Assuming for the moment that geometric differences between datasets can be addressed, has anyone considered how one would go about using a whole image dataset as an independent variable in, say, 3dRegAna, or what the pitfalls of this approach might be?

This would allow one to address issues like, "does the degree of gray matter density affect the level of activation?" or "can white matter density predict diffusion anisotropy?" on a voxel by voxel basis across populations of subjects.

I'm looking for any kind of comment at this point. Has anyone started doing this in ways other than a region of interest approach?

thanks,

jim