As far as I know, the result of fMRI data after AROMA denoised is only in NIFTI format, while the output of CIFTI data in fs_LR_32k space is not based on the data after AROMA denoising. So if I want to analyze the CIFTI data that has been denoised by AROMA, what should I do? Thanks a lot for help.
Hi @Guoyuan and welcome to Neurostars!,
In whatever software you are using to postprocess your data, you can include the AROMA regressors in the first-level model / denoising step. This would be equivalent to “aggressive AROMA denoising”.
If you want to use the nonaggressive denoising method, you could use something like ciftify or connectome workbench command
volume-to-surface-mapping. However, I would imagine that it might be better to project unsmoothed data to surface (fmriprep AROMA outputs are already smoothed), and then smooth on the surface. So you can take this code from the fMRIPrep documentation to nonaggressively noise one of your unsmoothed outputs, and proceed from there:
fsl_regfilt -i sub-<subject_label>_task-<task_id>_space-T1w_desc-preproc_bold.nii.gz \ -f $(cat sub-<subject_label>_task-<task_id>_AROMAnoiseICs.csv) \ -d sub-<subject_label>_task-<task_id>_desc-MELODIC_mixing.tsv \ -o sub-<subject_label>_task-<task_id>_space-T1w_desc-AROMAnonaggr_bold.nii.gz
Thank you very much for your quick reply. I think you are right. Smoothing really needs to be done in the surface space, so as to reduce the inappropriate correlation of the signal caused by partial volume effect. I will do the regression on the unsmoothed NIFTI data with AROMA variable, and use workbench to project NIFTI file to fs_LR32k space. Thanks again!