Dear Freesurfers, As freesurfer mri_glmfit-sim is based on AlphaSim & FSL randomise, I would like to know if the results I’m getting are affected by recently found bugs/violations in RFT assumptions, e.g. http://www.pnas.org/content/early/2016/06/27/1602413113.full.pdf.
Thank you in advance! Lauri Tuominen
There were bugs in AlphaSim (not FSL randomise); these do not affect FS because FS is an independent implementation.
FS does not use random field theory (RFT) but monte carlo simulation. The MC simulation does assume that the smoothing kernel is gaussian, so violations of that assumption will result in inaccurate results from the MC simulation. The PNAS paper found that the actual kernel had a heavier tail than gaussian resulting in larger clusters than expected and so more false positives.
I am in the process of doing similar analyses on thickness data. I'll will report when I have finalized them.
Dear Freesurfers, As freesurfer mri_glmfit-sim is based on AlphaSim & FSL randomise, I would like to know if the results Iâm getting are affected by recently found bugs/violations in RFT assumptions, e.g. http://www.pnas.org/content/early/2016/06/27/1602413113.full.pdf.
Thank you in advance! Lauri Tuominen
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
greve＠nmr.mgh.harvard.edu -
Lauri Tuominen