I am using melodic ICA to identify and remove noise components (using Feat v.6 pre-stats). I am currently re-running ICA on my raw data in order to apply b0 unwarping. I am now seeing many (10-20) new components that were not present in my initial ICA without b0 unwarping. All of these new components look pretty similar and show activity in areas with high susceptibility artifacts, so I am attributing these components to the unwarping. One such component can be viewed here: http://bit.ly/UidYO7. I am wondering whether or not to treat these components as noise and remove them before running lower-level Feats.
Thanks for any input on this.
Joe
|