Dear Ilana,
> Thanks, Jesper! I see what you mean about the misregistration between the repeated FA images. And I can appreciate that these differences should not substantially affect higher level analyses that involve additional registration steps (e.g., to standard space).
>
> I also made a quick genu mask and found that the mean FA values were less different for subsequent runs with nvoxhp=5000 (0.0014) versus nvoxhp=1000 (0.0065). So the difference in FA values is reduced when taking an average, as would be expected if there are slight misregistrations between images. Would you recommend reducing the variability by using nvoxhp=5000?
as I indicated I hope to release en new version very soon where this variability is anyway reduced.
The only disadvantage of a larger --nvoxhp is a longer execution time, so it never hurts.
Jesper
>
> Many thanks again,
> Ilana
|