Hi everyone,
I am running a set of GLM analyses. So far, I have addressed the issue of subjects' motion with the following procedure:
- I did not include subjects with visible motion artifacts in functional images
- I censored TRs with motion greater than 0.3mm (voxel size is 3mm3) and I included motion parameter estimates as regressors of no interest
However, I have realized that this results in censoring quite a lot TRs, which might eventually compromise the quality of the analysis. For most subjects, I excluded between 0 and 3% of the TRs, but I had to exclude as much as 10% in some subjects.
What is the limit above which a subject must be dropped? Could someone point me toward useful references?
Thank you!
Guillaume