So, I want to do some resting state functional connectivity work in the Harvard-Oxford atlas space, but I want to select my regions of interest by first modelling my task-based data in this space and then look at the N areas with most significant activation (largest betas). I’m using nipype and have fully set up 1st level analysis using FSL and it’s currently running now, but it’s currently running in the MNI space and I’m not sure this is the best thing to do.
If I run it in the MNI space, how would I then select the atlas regions with most activation? Simply projecting activations onto the atlas regions and then looking at average beta coeffs? Could I instead save lots of computing time by first moving to the atlas space and then running the GLM?
Either way, I actually don’t know how to move to atlas space using nipype. Is there a way to do this simply? I tried using some nilearn packages some months ago but they seem riddled with bugs at this stage in development…
I found TemplateFlow but as far as I can tell there is no instruction on how to actually use this package or what it does…