Hi Robie,
Note that I do not generally have to actually
worry about trimming subjects, but the proc
fields that seem important are perhaps:
- average censored motion: how much motion remains even after censoring?
- TRs censored (or the fraction)
- fraction TRs censored: per stim class
- and possibly GCOR
The TRs censored shows how much was lost,
but it is helpful to know how much motion there
was in the remaining TRs. The fraction per stim
class is important to know in case some class
is left with insufficient data. And GCOR might
be high for subjects with strong motion or
physiological contamination (e.g. from breathing).
The TSNR average should be somewhat inversely
scaled to motion. If that is low, there is usually a
more concrete field to point to.
It is not necessary to separate outliers from motion
censored TRs. They both indicate motion.
- rick