I’m running a first-level analysis on task-based fMRI data in SPM via a MATLAB script. Preprocessing was done with fMRIPrep. I’m currently including the six motion regressors converted from fMRIPrep with Remi Gau’s script (https://gist.github.com/Remi-Gau/06ace1387dcf9359f6abab663bceff3 — thanks for providing it).
I’d now like to add motion censoring based on an FD threshold, but I’m uncertain about the practical implementation in SPM12. Can I simply add a column to the motion-regressor file with 0 for kept frames and 1 for censored frames, or does SPM expect a different file, specific header or similar?
Any pointers on the recommended approach (and any SPM-specific caveats) would be much appreciated. Thank you!
Hey Dustin, short answer: yes. To my knowledge, you can simply append a series of binary spike regressors (e.g., from the fmriprep confounds file) to your matrix of motion parameters, save it all in a txt file, and then load that into your first-level GLM under the ‘multiple regressors’ option.
Since you’re using SPM, I might also recommend looking into Jorn Diedrichsen’s robust weighted least squares (rwls) toolbox as an alternative approach to ‘censoring’ noisy frames: Robust Weighted Least Squares Estimation
Hey Tyler,
thank you very much for the confirmation. This is very helpful! Also thank you for the hint towards the rwls toolbox. That might be a great fit for what I need for the first level analysis of my data.