Greetings FSL experts/users, In regards to using MELODIC ICA for strictly for denoising purposes, how important is the spatial smoothing kernel size for this process? I've been using a smoothing kernel of 0 which provides the most components through ICA, and thus I was under the assumption that this was in a sense 'safer' in regards to separating noise components from those one would want to keep. I just wanted to ask if there is any recommendation in regards to spatial smoothing, and if using a size of 0 would be acceptable (and thus apply spatial smoothing later on, on the denoised data set)? Thank you. Cheers, Paul