Hello FSL'ers,
I am working with resting state fMRI data and am trying to regress out WM and CSF as nuisance variables on AROMA output, prior to using group MELODIC. I found the following post on the listserve archives and have a few clarifying questions:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A3=ind1509&L=FSL&E=base64&P=376400&B=--_000_D20CE56E3615Ddanielyjyangyaleedu_&T=text%2Fhtml;%20charset=utf-8
I am guessing that the bold blue lettering are corrections made by Christian to the OP's post (see link above). According to the post:
fslmaths WM_mask_in_highres_space -ero WM_mask_in_highres_space_eroded
flirt -in WM_mask_in_highres_space_eroded \
-applyxfm \
-init rest.feat/reg/highres2example_func.mat\
-ref example_func.nii.gz \
-out WM_in_func_space
(I think this should be WM_mask_in_func_space to match what is used in fslmeants below)
fslmeants -i denoised_func_data.nii.gz -m WM_mask_in_func_space --no_bin -o WM_in_func_bin_timeseries
Here are my questions:
Question 1) is the "WM_mask_in_highres_space" the same as the t1_pve_2.nii.gz (for WM) returned by FAST?
Question 2) which AROMA denoised data do I use, the original denoised data or that warped to the standardized space with or without the highpass filtering applied? To make sense of what I did, first, the FEAT GUI was used to register the resting state images to the t1 images, this gave the /reg folder inside the .feat directory. This .feat directory was passed in to AROMA which gave the denoised_func_data_aggr.nii.gz file. Then, as was explained in the ICA practical (near the bottom), I warped the data in to the standard space via the script
applywarp -r ${j}/reg/standard.nii.gz -i ${j}/ICA_AROMA_aggr/filtered_func_data_clean.nii.gz -o ${j}/ICA_AROMA_aggr/ filtered_func_data_clean_standard.nii.gz --premat=${j}/reg/example_func2highres.mat -w ${j}/reg/highres2standard_warp.nii.gz
So we have the first .feat directory (containing filtered_func_data.nii.gz) which is passed in to AROMA in which we get the output folder named ICA_AROMA_aggr inside of which has filtered_func_data_clean.nii.gz and after applywarp is applied to the clean data has filtered_func_data_clean_standard.nii.gz. Then, as the AROMA manual suggested, temporal highpass filtering (THP) is done after AROMA, so we then have filtered_func_data_clean_standard_THP.nii.gz. While I would guess it is the latter image, which image should actually be used as input to fslmeants?
Now after we are finished with fslmeants we do the exact same methodology for CSF and get the CSF_in_func_bin_timeseries which then I combine the two text files in to the nuisance_timeseries text file:
paste WM_in_func_bin_timeseries CSF_in_func_bin_timeseries > nuisance_timeseries
Question 3) which "denoised_func_data" do I use in fslmaths and fsl_glm in the following code?
fslmaths denoised_func_data –Tmean tempMean
fsl_glm –i denoised_func_data –d nuisance_timeseries –demean –res_out=residual
Question 4) The OP seems to be doing temporal highpass filtering in his last step via fslmaths. Is there an advantage to the way he is doing it vs. calling FEAT again and selecting nothing but the temporal highpass filtering button and running?
Question 5) A bit of an aside question on the warping script (applywarp above) as seen in the ICA Practical. Unlike the applywarp script in the ICA practical, I used the " filtered_func_data_clean.nii.gz " from the aggressive run of AROMA which was output in the ICA_AROMA_aggr/ folder. Is this correct? This is different from using the filtered_func_data_clean.nii.gz directly in ${j} as was shown in the practical as this image did not exist in this location.
Question 6) At what point of the image processing pipeline (including AROMA) should nuisance regression such as WM and CSF be done?
Thanks,
FSLMonk
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
|