Dear SPM-Experts,
we have conducted a rather complex multimodal study using among others fMRI, which includes a clinical sample. Right now we are in the review process with the paper from this dataset. During the initial analysis we corrected for motion during preprocessing using the "Realign and Unwarp" function and later we added the motion rotation parameters as regressors to the model. During the review process one of the reviewers was concerned that the additional motion threshold we applied was too lenient and suggested a scan-to-scan motion threshold of half the voxel-size (1.5 mm), which is truly quite common in clinical datasets, I think. The reviewer additionally asked us to consider the use of data scrubbing techniques to potentially deal with data showing too much motion. We applied the suggested threshold and found that the rotation parameters only exceeded this threshold in 18 scans out of the almost 80 000 scans we acquired in total. This amounts to 0.02% of the data and the maximum within one participant was 0.33% of the scans. No scan-to-scan movement exceeded 3.5 mm. I have two questions regrading this:
1. Considering the amount of scans affected and the methods already applied, is it necessary to employ additional techniques to correct for motion, such as data scrubbing?
2. If it is necessary to employ a form of data scrubbing, which method would be the most economic method to do this? (As stated, we had already finished the analysis)
Thank you for your help!
Best,
Paul
|