help
help > RE: Data in ROI REX files
Dec 21, 2018 09:12 PM | Alfonso Nieto-Castanon - Boston University
RE: Data in ROI REX files
Dear Allison,
Regarding (1), yes, exactly, the values in those REX files are just the raw values (before Denoising)
Regarding (2), yes, exactly, as well. Five is the default number of components used by aCompCor in CONN denoising pipeline, but of course you may modify that in the Denoising tab by selecting any of those two components and changing the "confound dimensions" entry before running the Denoising step)
And regarding (3), not explicitly. The files named rp_*.txt generated during realignment will contain the 6 motion parameters (three translation and three rotation parameters), but the additional 6 parameters (consisting on the first-order temporal derivatives of the above) are computed on the fly during the Denoising step but never explicitly stored in an output file. If you want you may explicitly compute/create new files with those 12 parameters using something like:
f = conn_module('get','l1covariates','realignment');
cellfun(@(x)spm_save(conn_prepend('e',x,'.csv'),[load(x) gradient(load(x)')']),[f{:}]);
(this will create a new set of files -one for each subject and session- named erp_*.csv in the same folders as the original rp_*.txt realignment files but containing the 12 motion parameters instead)
Hope this helps
Alfonso
Originally posted by allison_shapiro:
Regarding (1), yes, exactly, the values in those REX files are just the raw values (before Denoising)
Regarding (2), yes, exactly, as well. Five is the default number of components used by aCompCor in CONN denoising pipeline, but of course you may modify that in the Denoising tab by selecting any of those two components and changing the "confound dimensions" entry before running the Denoising step)
And regarding (3), not explicitly. The files named rp_*.txt generated during realignment will contain the 6 motion parameters (three translation and three rotation parameters), but the additional 6 parameters (consisting on the first-order temporal derivatives of the above) are computed on the fly during the Denoising step but never explicitly stored in an output file. If you want you may explicitly compute/create new files with those 12 parameters using something like:
f = conn_module('get','l1covariates','realignment');
cellfun(@(x)spm_save(conn_prepend('e',x,'.csv'),[load(x) gradient(load(x)')']),[f{:}]);
(this will create a new set of files -one for each subject and session- named erp_*.csv in the same folders as the original rp_*.txt realignment files but containing the 12 motion parameters instead)
Hope this helps
Alfonso
Originally posted by allison_shapiro:
Dear CONN Community,
I am working with the extracted BOLD signal data that is output in the REX files after the set-up step in CONN, and I am hoping to get some clarification on what data are in these files.
1) For any given ROI of interest, are the BOLD signal values in the REX file confound corrected (movement, CSF, WM), or are they just raw values? My assumption is that they are raw but need to confirm this.
2) In the WM and CSF ROI REX files there are 16 variables. The first appears to be the raw signal and the remaining 15 appear to be the CompCor values for the principle components. However, in the denoising step in CONN, both the WM and CSF show that only 5 confounding effects will be regressed. Does the CompCor method only use the first 5 principle components for denoising?
3) Is there any data file that is output after set-up that contains data for the 12 movement confounding variables? If so, where would I find this?
Any input or guidance is much appreciated! Thank you for such a great toolbox!
Best Wishes,
Allison
I am working with the extracted BOLD signal data that is output in the REX files after the set-up step in CONN, and I am hoping to get some clarification on what data are in these files.
1) For any given ROI of interest, are the BOLD signal values in the REX file confound corrected (movement, CSF, WM), or are they just raw values? My assumption is that they are raw but need to confirm this.
2) In the WM and CSF ROI REX files there are 16 variables. The first appears to be the raw signal and the remaining 15 appear to be the CompCor values for the principle components. However, in the denoising step in CONN, both the WM and CSF show that only 5 confounding effects will be regressed. Does the CompCor method only use the first 5 principle components for denoising?
3) Is there any data file that is output after set-up that contains data for the 12 movement confounding variables? If so, where would I find this?
Any input or guidance is much appreciated! Thank you for such a great toolbox!
Best Wishes,
Allison
Threaded View
| Title | Author | Date |
|---|---|---|
| allison_shapiro | Dec 17, 2018 | |
| Alfonso Nieto-Castanon | Dec 21, 2018 | |