Hello!

After running two sessions of data through the preprocessing pipeline for volume-based analyses (indirect normalization to MNI-space) when FieldMaps are available, we moved on to the 1st-level denoising step, but found what appears to be some kind of artifact in the proportion of variance explained in the BOLD signal by the total confounds. I understand that a lack of suprathresholded voxels only indicates that the contribution of the confounds is below 50% of the total variance in the BOLD signal, but there's a distinct band-like pattern to the loss of effect that consistently occurs in the superior-anterior region of the brain on both sessions for about 20% of our subjects (see attached screenshot of the two worst ones). When I increase the threshold, the very small effect turns to a complete loss of any effect of the confounds in those areas. I'm mostly concerned because it has such a consistent and distinct pattern. Is there something we should be doing differently in the preprocessing or change in the denoising settings, or is this something we don't need to worry about?

Thanks a ton for your help!
Hannah