help > Problem with WM, CSF components in repeat analysis
Showing 1-3 of 3 posts
Jul 28, 2020 04:07 PM | Adam Kaminski - Georgetown University
Problem with WM, CSF components in repeat analysis
Hello CONN community,
We ran through a preprocessing pipeline in CONN that included ART outlier detection only. We did this in December and then again in July with the same subjects and noticed some differences in the WM and CSF components (picture of an example subject attached). We think this led to different outcomes in the second level analysis. The CONN version was the same (18a). Does anyone know what might be causing this?
Thank you!
Adam
We ran through a preprocessing pipeline in CONN that included ART outlier detection only. We did this in December and then again in July with the same subjects and noticed some differences in the WM and CSF components (picture of an example subject attached). We think this led to different outcomes in the second level analysis. The CONN version was the same (18a). Does anyone know what might be causing this?
Thank you!
Adam
Aug 5, 2020 01:08 PM | Alfonso Nieto-Castanon - Boston University
RE: Problem with WM, CSF components in repeat analysis
Hi Adam,
I see no reasonable explanation for that, other than having manually modified between the two analyses the source of either your functional data or your WM or CSF masks (since neither the functional data nor the WM/CSF masks are really touched or modified in any way by ART). Is there any simple way you can check what functional data files and Gray/White/CSF ROIs you were using in the two cases?
Alfonso
Originally posted by Adam Kaminski:
I see no reasonable explanation for that, other than having manually modified between the two analyses the source of either your functional data or your WM or CSF masks (since neither the functional data nor the WM/CSF masks are really touched or modified in any way by ART). Is there any simple way you can check what functional data files and Gray/White/CSF ROIs you were using in the two cases?
Alfonso
Originally posted by Adam Kaminski:
Hello CONN community,
We ran through a preprocessing pipeline in CONN that included ART outlier detection only. We did this in December and then again in July with the same subjects and noticed some differences in the WM and CSF components (picture of an example subject attached). We think this led to different outcomes in the second level analysis. The CONN version was the same (18a). Does anyone know what might be causing this?
Thank you!
Adam
We ran through a preprocessing pipeline in CONN that included ART outlier detection only. We did this in December and then again in July with the same subjects and noticed some differences in the WM and CSF components (picture of an example subject attached). We think this led to different outcomes in the second level analysis. The CONN version was the same (18a). Does anyone know what might be causing this?
Thank you!
Adam
Aug 11, 2020 06:08 PM | Adam Kaminski - Georgetown University
RE: Problem with WM, CSF components in repeat analysis
Thank you for your response! I have checked the functional files
and ROIs and they match, but of course there may have been an error
somewhere. We will continue to investigate and I'll report back any
conclusions!
Adam
Adam