help
487 Subscribers 
help > Structural data and gray/white/CSF masks
Showing 1-2 of 2 posts
Nov 2, 2016 01:11 AM | Ruedeerat Keerativittayayut - Kochi university of technology
Structural data and gray/white/CSF masks
Dear Alfonso,
I have question about structural images that will use to derive the gray, white, and CSF mask. In my case I want to skip preprocessing processes because I already have SPM.mat and swu* functional images.
When I loaded my SPM.mat, i noticed that the program automatically loaded functional data into functional module, condition parameters (onset, duration) into conditions module,
WM/CSF/motion parameters into covariates module. So, I have to put structural and ROI module manually. My question is if I left blank at the grey matter, white matter, and CSF inside ROIs module, what type of structural image I should put. It should be normalized image, original image? If I put wBrain (normalized structural images), Is its resolution OK for performing segmentation to get GM WM CSF. The program will confuse with my WM and CSF parameters that already automatically loaded into the covariates module?
Thank you in advance.
Best Regards,
Fai
I have question about structural images that will use to derive the gray, white, and CSF mask. In my case I want to skip preprocessing processes because I already have SPM.mat and swu* functional images.
When I loaded my SPM.mat, i noticed that the program automatically loaded functional data into functional module, condition parameters (onset, duration) into conditions module,
WM/CSF/motion parameters into covariates module. So, I have to put structural and ROI module manually. My question is if I left blank at the grey matter, white matter, and CSF inside ROIs module, what type of structural image I should put. It should be normalized image, original image? If I put wBrain (normalized structural images), Is its resolution OK for performing segmentation to get GM WM CSF. The program will confuse with my WM and CSF parameters that already automatically loaded into the covariates module?
Thank you in advance.
Best Regards,
Fai
Nov 2, 2016 07:11 PM | Alfonso Nieto-Castanon - Boston University
RE: Structural data and gray/white/CSF masks
Dear Fai
If your SPM.mat files already defined WM/CSF/motion timeseries, then when importing these SPM.mat files in CONN those will appear as a new first-level covariate in CONN named 'SPMcovariates' containing all of those timeseries.
If you are planning to use those same timeseries in the Denoising step (e.g. instead of using the default aCompCor procedure in CONN consisting of segmenting your structural data and extracting PCA components from White/CSF areas), then it does not really matter what structural data you put into Setup.Structurals and what White/CSF masks you use in Setup.ROIs. When you get to the Denoising tab you would simply replace from the 'Confounds' list the default White and CSF effects there, and enter instead your SPMcovariates effect in that list.
If, on the other hand, you prefer to use the default aCompCor procedure in CONN then yes, enter in Setup.structural T1 volumes for your subjects (normalized if possible; if you do not have normalized volumes then enter your raw volumes and then use the Preprocessing button to run a "structural normalization&segmentation" step), and then enter in Setup.ROIs the normalized Gray/White/CSF masks for your subjects (if you do not have these already, then running the above "structural normalization&segmentation" step will automatically create these for you; or alternatively leaving these empty will automatically run a single "structural segmentation" step on these when needed)
If you are unsure, or want to try both methods, then follow the second set of instructions above to enter your Structural/ROI data, and then when you get to the Denoising tab you will be able to select which White/CSF timeseries you would like to use for denoising (i.e. the ones in the 'SPMcovariates' variable or the ones in the 'White" and "CSF" ROIs)
Hope this helps
Alfonso
Originally posted by Ruedeerat Keerativittayayut:
If your SPM.mat files already defined WM/CSF/motion timeseries, then when importing these SPM.mat files in CONN those will appear as a new first-level covariate in CONN named 'SPMcovariates' containing all of those timeseries.
If you are planning to use those same timeseries in the Denoising step (e.g. instead of using the default aCompCor procedure in CONN consisting of segmenting your structural data and extracting PCA components from White/CSF areas), then it does not really matter what structural data you put into Setup.Structurals and what White/CSF masks you use in Setup.ROIs. When you get to the Denoising tab you would simply replace from the 'Confounds' list the default White and CSF effects there, and enter instead your SPMcovariates effect in that list.
If, on the other hand, you prefer to use the default aCompCor procedure in CONN then yes, enter in Setup.structural T1 volumes for your subjects (normalized if possible; if you do not have normalized volumes then enter your raw volumes and then use the Preprocessing button to run a "structural normalization&segmentation" step), and then enter in Setup.ROIs the normalized Gray/White/CSF masks for your subjects (if you do not have these already, then running the above "structural normalization&segmentation" step will automatically create these for you; or alternatively leaving these empty will automatically run a single "structural segmentation" step on these when needed)
If you are unsure, or want to try both methods, then follow the second set of instructions above to enter your Structural/ROI data, and then when you get to the Denoising tab you will be able to select which White/CSF timeseries you would like to use for denoising (i.e. the ones in the 'SPMcovariates' variable or the ones in the 'White" and "CSF" ROIs)
Hope this helps
Alfonso
Originally posted by Ruedeerat Keerativittayayut:
Dear Alfonso,
I have question about structural images that will use to derive the gray, white, and CSF mask. In my case I want to skip preprocessing processes because I already have SPM.mat and swu* functional images.
When I loaded my SPM.mat, i noticed that the program automatically loaded functional data into functional module, condition parameters (onset, duration) into conditions module,
WM/CSF/motion parameters into covariates module. So, I have to put structural and ROI module manually. My question is if I left blank at the grey matter, white matter, and CSF inside ROIs module, what type of structural image I should put. It should be normalized image, original image? If I put wBrain (normalized structural images), Is its resolution OK for performing segmentation to get GM WM CSF. The program will confuse with my WM and CSF parameters that already automatically loaded into the covariates module?
Thank you in advance.
Best Regards,
Fai
I have question about structural images that will use to derive the gray, white, and CSF mask. In my case I want to skip preprocessing processes because I already have SPM.mat and swu* functional images.
When I loaded my SPM.mat, i noticed that the program automatically loaded functional data into functional module, condition parameters (onset, duration) into conditions module,
WM/CSF/motion parameters into covariates module. So, I have to put structural and ROI module manually. My question is if I left blank at the grey matter, white matter, and CSF inside ROIs module, what type of structural image I should put. It should be normalized image, original image? If I put wBrain (normalized structural images), Is its resolution OK for performing segmentation to get GM WM CSF. The program will confuse with my WM and CSF parameters that already automatically loaded into the covariates module?
Thank you in advance.
Best Regards,
Fai