Hello Joyce,
Which version of FSL are you using? Randomise 2.1 is quite an old version. We always recommend updating to the latest version of FSL ( currently 5.0.6 which contains randomise 2.9 ).
Kind Regards
Matthew
> Dear Anderson,
>
> Thank you very much for your clarification. You are absolute right, I checked the both *p and *corrp results, where *p statistic results are more significant than the FWE corrected results that is understandable. I now understand the difference between randomise results and results of a random extracted voxel in SPSS. But are you saying it is possible not to have raw statistical values in massive brain regions in *tstat maps, even if the PBVC revealed significant differences between two groups? In my randomise *tstat results
>
> I once ran the same data using the same matrix and contrast one year ago. I had completely different results compared to the randomise results I have now. I checked randomise versions which are the same Randomise 2.1. I used the same command line which I showed on the first poster. The first randomise (I ran one year ago) results showed regional brain atrophy in the bilateral frontal and temporal lobe, where the clusters were normally bigger than 300 voxel. But the second randomise results which I ran recently, showed regional brain differences between diagnosis groups in the right frontal and temporal lobe with cluster size no more than 100 voxel. How does this happen? Is there a possible way to further exam what caused these two different results? Looking forward to your reply.
>
> BW,
> Joyce
|