It may be the case that the result are not significant. If you have an apriori region then you can analyze only in that region rather than doing a whole-brain analysis. That will make the result more significant.
On 10/9/16 4:08 PM, Karamfil Bahchevanov wrote:
Hello Freesurfer experts,
I'm really struggling with my study. I have a 27 scans that I analyze in QDEC. Scans are from Siemens machine, mprage protocol, TR1620, TE 4.3, FOV 240*256, acq. matrix 232*256, slice thickness 0,9mm. After I enter model and have some initial significant clusters (threshold 2 - 5) I run multiple comparisons, and none have survived the procedures, no matter method or significance - FDR (0.1-0.05-0.01) or Monte-Carlo(0.05-0.01-0.01). This repeats no matter of the tested variables and the initial significancy and size of the clusters. I'm really confused of that and could not find an explanation, except that my study is really non-significant. However, the significant results I get are in the regions that I expect to be, that are reported previously and they also explain my results. If they happen to survive, but that never is the case (In fact one survived, but I entered wrong variables and model was wrong). What could be the reason? Outliers? Really non-significant results (which I doubt as I said already)? Should I recheck the quality of the processed scans? I also made my own average, hoping to improve things somehow, but QDEC doesn't allow me to run multiple comparisons with it. Anyway, is it mistake to discuss the results when they don't survive simulation, as they fit with initial hipothesis and what I'm testing. I have to point that there are only few significant clusters, that could not be explained with what I'm testing and tested population, so in fact the "noise" is very small. Any help will be of great value,
Thanks
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer