Hi,

Thanks a lot for sharing these clearly-written scripts for fc analysis.

I have one question regarding the dual regression.
When we use the metaICA (or our own group ICA maps) as spatial regressor to extract time courses, do we get the mean of everything (both positive and negative values within the ICA maps) or only the positive of it? Does it make sense to threshold the metaICA (or group's ICA maps) and take own the mean time courses from those regions? Any advice on if we are interested in the anti-correlations (the negatives), does it make sense to thrshold the metaICA and invert the negative image?

Thanks a lot in advance for any inputs.

Best,
Monica

Hi Monica,

Yes. DR can give both pos and neg values in terms of your 4D timeseries. The way of getting the avg timeseries based on ROIs derived from metaICA/gICA maps you described, to me, sounds ok. I am not very sure the meaning that you want to invert the negative image, sorry. Could you give some details about it?

Best,
Xi-Nian

Hi Xi-Nian,

Thanks a lot for your reposnse!! To clarify my questions, because there are some "negative" shown in the group ICA we got, although not a lot. So, I am wondering, by takeing the average timecourse using the entire ICA spatial map as mask, is it going to average out the time courses from both positive and negative voxels? How is that going to affect the results? So, I am wondering if it would make more sense to threshold the ICA (say Z>2.3 or whatever value), and use the thresholded ICA as spatial regressor? Please forgive me if I am asking something stupid, but I am confused about this...

Thanks a lot!!

Best,
Monica