Hi,
> I tried running TBSS for the first time, and i want to make sure I understand the results correctly.
>
> I got two contrast files, tbss_tfce_corrp_tstat1 and tbss_tfce_corrp_tstat2, corresponding to patients>controls and controls>patients, respectively. When i open the second contrast I get areas of significant difference indicated in red ( 0.95-1, p<0.05). As far as I understand, these areas indicate regions of reduced FA in the patient group ( since this is the control>patient contrast).
That's correct.
> When I open the first contrast ( patient>controls), no areas of significance are seen ( just the green skeleton). This makes sense if I described the results of the second contrast correctly.
Yes, this makes sense as you do have the correction description.
> Could i look at the results by overlaying the two contrasts? i tried doing that, and it seems to me like the areas of significance expended ( for the second contrast, controls>patients, more areas of significance were detected in blue-light blue, while for the first contrast, patients>controls, small spots of significance were detected in reddish, though no areas of difference were observed when uploading this contrast by itself. I a, not sure how to read these results....) Does red always indicate reduced FA and blue increased FA?
I'm not sure why you are doing this. According to your description above you have no significant results in contrast 1, so there's nothing to be gained by investigating this further (unless you are looking for sub-threshold results that did not reach statistical significance, but I'm not sure why you'd do this as you already have a perfectly interpretable and significant finding in contrast 2 on its own). I would just interpret the significant findings that you have from contrast 2 and leave it at that.
> And lastly, I ran randomise with -T2 and -n=5000. Some sources are confusing in that they use 500 instead 5000 (but maybe with T2 its ok?), or use T when T2 should be used for TBSS.
Using 5000 gives you more accurate p-values, so you are better off with this and there's no need to change it.
The --T2 option is the recommended one for TBSS as it acts on a skeleton (a reduced subset of the 3D data).
All the best,
Mark
> Thank you in advance,
> Anahit
|