Hi Steve,
Thanks for the help! Based off what you said I had two follow up questions:
1) When reading the melodic preprocessing steps from FEAT (http://fsl.fmrib.ox.ac.uk/fslcourse/lectures/feat1_part1.pdf) I saw that there is a grand intensity normalization step. I have seen this used in tasked based experiments, however would this also be a valid preprocessing step for Resting state ICA analysis, and if so do you see a major difference in results?
2) For global signal regression I have read the pro's and con's . For example if there is subject motion, the effects are exacerbated by increasing near range connectivity and decreasing far away connectivity (or introduce negative correlations). One of the pros is that helps when doing a two group comparison and may increase sensitivity to detecting changes between groups. Most of these studies have used seed based analysis. Do you suspect these pros and cons findings hold true for ICA analysis?
Thanks,
Ajay
|