[color=#000000]Hi Chaleece,[/color]
[color=#000000]Yes, that definitely looks like residual noise in the seed-to-voxel correlation maps. Just to clarify, could you please comment on whether the variance/size of the voxel-to-voxel histogram for that subject (displayed in the figures that you attached) looks similar or larger than that expected from other subjects in your experiment (e.g. use the 'show all' button to compare these histograms across all subjects)? (the reason I ask is because to my eye this denoised histogram looks "wider" than what I would expect, this may reflect simply relatively short scanning time -e.g. if that looks similar across all subjects- or it may reflect an issue with this particuar subject -e.g. if that looks narrower for other subjects-). In general, potential things to try to make the Denoising step more conservative would be to increase the number of dimensions in the 'WhiteMatter' and 'CSF' confounding effects, and/or using more conservative ART thresholds. Another venue in this case would be to try to identify why this particular subject may show higher residual noise than expected (e.g. perhaps this reflects an issue during preprocessing? was this subject moving significantly more than others? was this movement constrained to one particular session?) and try to address those issues directly if possible. Last keep in mind that if the degrees of freedom in your dataset are relatively low across the board then making the Denoising step more conservative will push those degrees of freedom even lower, negatively affecting the quality of the data for other subjects, so it may be more effective to simply disregard this subject (or some subset of runs/sessions for this subject) from the analyses.[/color]
[color=#000000]Hope this helps[/color]
[color=#000000]Alfonso[/color]
[i]Originally posted by Chaleece Sandberg:[/i][quote]Hello good people!
I am analyzing resting state data and using the default pipeline.
I have one subject for which there are outliers from art, but the scrubbing does not show any % BOLD explained. In the first level results, it looks as if there are "motion-contaminated" results. I am attaching a screenshot.
1. Why might scrubbing not help with this subject?
2. What can I do to improve the denoising for this individual?
I have been perusing the forum, but can't seem to find an answer that fits.
Any advice is greatly appreciated![/quote]
[color=#000000]Yes, that definitely looks like residual noise in the seed-to-voxel correlation maps. Just to clarify, could you please comment on whether the variance/size of the voxel-to-voxel histogram for that subject (displayed in the figures that you attached) looks similar or larger than that expected from other subjects in your experiment (e.g. use the 'show all' button to compare these histograms across all subjects)? (the reason I ask is because to my eye this denoised histogram looks "wider" than what I would expect, this may reflect simply relatively short scanning time -e.g. if that looks similar across all subjects- or it may reflect an issue with this particuar subject -e.g. if that looks narrower for other subjects-). In general, potential things to try to make the Denoising step more conservative would be to increase the number of dimensions in the 'WhiteMatter' and 'CSF' confounding effects, and/or using more conservative ART thresholds. Another venue in this case would be to try to identify why this particular subject may show higher residual noise than expected (e.g. perhaps this reflects an issue during preprocessing? was this subject moving significantly more than others? was this movement constrained to one particular session?) and try to address those issues directly if possible. Last keep in mind that if the degrees of freedom in your dataset are relatively low across the board then making the Denoising step more conservative will push those degrees of freedom even lower, negatively affecting the quality of the data for other subjects, so it may be more effective to simply disregard this subject (or some subset of runs/sessions for this subject) from the analyses.[/color]
[color=#000000]Hope this helps[/color]
[color=#000000]Alfonso[/color]
[i]Originally posted by Chaleece Sandberg:[/i][quote]Hello good people!
I am analyzing resting state data and using the default pipeline.
I have one subject for which there are outliers from art, but the scrubbing does not show any % BOLD explained. In the first level results, it looks as if there are "motion-contaminated" results. I am attaching a screenshot.
1. Why might scrubbing not help with this subject?
2. What can I do to improve the denoising for this individual?
I have been perusing the forum, but can't seem to find an answer that fits.
Any advice is greatly appreciated![/quote]