Hi,
> I am doing a Two sample TTest between the FA images of
> two groups using TBSS with the TFCE option (--T2 in
> randomise). How can I further threshold the resulted image
> "tbss_tfce_corrp_tstat" by cluster size? And how
> can I extract the most statistically significant cluster? I
> am planning to deproject it back to the native space of each
> subject for further analyses but there are too many clusters
> in the 'tbss_tfce_corrp_tstat' file. Detail steps
> here would be extremely appreciated!
You can always "push" the correction further than p<0.05 (with P<0.01 or even P<0.001 for instance). To get the clusters, use the cluster command, e.g.: cluster -i my_tfce_corrp_tstat* -t 0.95 (or 0.99 etc.) -o my_significant_clusters.txt.
If by "further analysis" you mean doing some tractography, please bear in mind that now probtrackx accepts masks in the standard space (as just mentioned by Matt Glasser), so you would just need to deproject the results back in the centre of the tract of each subject in the standard space (tbss_deproject 1).
Cheers,
Gwenaelle
|