I am using the CONN toolbox to preprocess and denoise the MRI data of ABCD study. I am attaching the fMRI data that I obtained after preprocessing and denoising. This nifti data was converted from "DATA_SubjectXXX_Condition001.mat" under the folder /results/preprocessing. There are two problems that I can observe: (1) The signal did not normalize to percentage although I set it to percent signal change. (2) The upper part of the brain is missing. I observed other subjects can miss the back part of the brain. Could you please diagnose what kind of mistake I might make during the preprocessing or the denoising step?
During the procedure, I chose the default setup except for "slice order: do not know" during preprocessing and skipping scrubbing during denoising. Please note for each participant, I processed two fMRI scans. So CONN toolbox concatenated the two runs while processing them. I am not sure if this matters. Thanks for the help!
Hi Ru Zhang
I am not sure, (1) I imagine may be related to the mean signal having been removed from your data during denoising (so the average BOLD signal is 0 instead of 100), while (2) might indicate some problem during the normalization step (for that I would probably suggest to run some of the QA plots in CONN to look at the quality of the normalization results).
Another suggestion would be to select in the Setup.Options the option to create confound-corrected timeseries, as that will create denoised .nii files directly (one file per session instead of concatenating across sessions, with filenames prepending the letter 'd' to the preprocessed filenames, e.g. dswaurest.nii files) that are typically better suited for exporting to other software packages.
Hope this helps
Alfonso
Originally posted by Ru Zhang:
I am using the CONN toolbox to preprocess and denoise the MRI data of ABCD study. I am attaching the fMRI data that I obtained after preprocessing and denoising. This nifti data was converted from "DATA_SubjectXXX_Condition001.mat" under the folder /results/preprocessing. There are two problems that I can observe: (1) The signal did not normalize to percentage although I set it to percent signal change. (2) The upper part of the brain is missing. I observed other subjects can miss the back part of the brain. Could you please diagnose what kind of mistake I might make during the preprocessing or the denoising step?
During the procedure, I chose the default setup except for "slice order: do not know" during preprocessing and skipping scrubbing during denoising. Please note for each participant, I processed two fMRI scans. So CONN toolbox concatenated the two runs while processing them. I am not sure if this matters. Thanks for the help!