Sorry for beeing unclear!

Ok, so the newest version runs, when using anaticor, both 3dD and 3dREMLfit? That is what our version does, I get both the stats file and the stats.REML file.

So the question is if it is worth the extra time and the loss of DOF to run BP filtering. You have me pretty convinced, I'm not a huge fan of BP either. But we did see improvments a while back when we ran some optimization/QA with different setitngs.

Do I got this right? The Nyqvist frequency is 1 / (TR*2) = 0.55 which means that our upper limit would be 0.01 0.55. We use (for a reason I don't really remember) 1 / (TR *3) which is 0.37. So we could increase it even more (and use up even more DOF)? I think the logic we use is that we can't detect changes that are faster than 3 * TR but perhaps that is wrong and the true max frequency is 0.55?

Thanks!



Edited 1 time(s). Last edit at 06/26/2017 04:34PM by Robin.