The residuals are something in the FMRI signal that you could not explain with those regressors you specified in the model. So if you have resting state data under two different conditions, it does not make sense to obtain the contrast of the two residual time series. Instead, just treat them the same way (residuals are just residuals, and presumably/ideally have nothing to do with the conditions). You could estimate FWHM values from each residual series separately, and then use the average for multiple testing correction. Or you could concatenate the two residual time series before estimating the FWHM values.
Gang
Edited 1 time(s). Last edit at 03/25/2015 12:40PM by Gang.