Running the analysis without OD changes a lot, more than I've noticed when running single group third levels. Seems like something weird is happening.
Also, the other question I had pertained to the 4D prob_outlier images. My understanding is that each volume in the time series gives the voxelwise probability of being an outlier. It seems strange to me then to find many of the volumes in a prob_outlier image to contain identical patterns of high probability. I would have thought that these volumes would be more idiosyncratic and heterogeneous.
thanks,
david
|