Print

Print


Dear Maxwell,

>I am trying to do a longitudinal study using repeated measures ie
>perform fmri study in 12 patients at 2wks, 4 wks, 6 wks, 8 wks, 12wks,
>6 months and 1 year. I will collect motor recovry score(number 
>ranging from 1-100) at each
>of these time points. I am interested in correlations between fmri
>activation and the motor score. I would like to assess if individual
>activations change with time and whether these changes correlate with
>motor score. What is the best way to analyze the data at individual
>level and group level.

I would use a three-stage analysis; at the between-scan, between-
repetition and between-subject levels.  This entails estimating
the session-specific motor activations for each repetition and subject.
Then take the contrast for motor-related responses to a second level to
test for within-subject effects of time and or recovery; note these will
be confounded, so do one at a time. Use age etc as a confound.
Finally, take the contrasts testing for an effect of time (or recovery)
to a third between-subject level, using a one-sample t-test.

I hope this helps - Karl