help > Default number of dimensions for White matter and CSF Denoising confounds in batch scripts?
Showing 1-4 of 4 posts
Jun 24, 2022 07:06 PM | Karl Lerud
Default number of dimensions for White matter and CSF Denoising confounds in batch scripts?
In my batch scripts, I usually specify denoising confounds like
this:
BATCH.Denoising.confounds.names={'White Matter' 'CSF' 'realignment' 'scrubbing'};
But I noticed that when I open those projects in the GUI, in the Denoising tab, it says (16P) by White matter and CSF, which is the same as specifying Inf number of components/dimensions right? After looking around, 5 seems to be a much more common number for White matter and CSF confounds, and I think it defaults to 5 if you make the whole project in the GUI, is that right?
I believe I figured out how to specify this in the script:
BATCH.Denoising.confounds.names={'White Matter' 'CSF' 'realignment' 'scrubbing'};
BATCH.Denoising.confounds.dimensions={5 5 Inf Inf};
When I do this, it seems to only utilize 5 dimensions for White matter and CSF, which is what I intended. And the results are very different after only changing this aspect of the analysis, generally with many Fisher correlations going up. I wanted to check:
1) Is 5 a good number to stay with for this, and/or how would one choose this number,
2) Is defaulting to 16/Inf expected behavior for a batch script without specifying the number of dimensions, and
3) Am I specifying this correctly in my script with that line?
Thanks, best,
Karl
BATCH.Denoising.confounds.names={'White Matter' 'CSF' 'realignment' 'scrubbing'};
But I noticed that when I open those projects in the GUI, in the Denoising tab, it says (16P) by White matter and CSF, which is the same as specifying Inf number of components/dimensions right? After looking around, 5 seems to be a much more common number for White matter and CSF confounds, and I think it defaults to 5 if you make the whole project in the GUI, is that right?
I believe I figured out how to specify this in the script:
BATCH.Denoising.confounds.names={'White Matter' 'CSF' 'realignment' 'scrubbing'};
BATCH.Denoising.confounds.dimensions={5 5 Inf Inf};
When I do this, it seems to only utilize 5 dimensions for White matter and CSF, which is what I intended. And the results are very different after only changing this aspect of the analysis, generally with many Fisher correlations going up. I wanted to check:
1) Is 5 a good number to stay with for this, and/or how would one choose this number,
2) Is defaulting to 16/Inf expected behavior for a batch script without specifying the number of dimensions, and
3) Am I specifying this correctly in my script with that line?
Thanks, best,
Karl
Jul 15, 2022 08:07 PM | Karl Lerud
RE: Default number of dimensions for White matter and CSF Denoising confounds in batch scripts?
Hi Alfonso, can I bump my question here, about the number of
parameters/dimensions of denoising confounds? Thanks,
Karl
Karl
Jan 14, 2025 07:01 PM | Christopher Hyatt - Olin Neuropsychiatry Research Center
RE: Default number of dimensions for White matter and CSF Denoising confounds in batch scripts?
Anyone figure out what White Matter (16P) and CSF (16P) refer to? Is this 16 aCompCor (PCA-derived) components, or WM/CSF signals from 16 (random?) locations within WM/CSF?
Jan 14, 2025 07:01 PM | Christopher Hyatt - Olin Neuropsychiatry Research Center
RE: Default number of dimensions for White matter and CSF Denoising confounds in batch scripts?
After RTFM ;), I printed out the 'Methods' and the 16P refers to 16 aCompCor components each for WM and CSF.