Hello AFNI minds,
I am interested in going forward with 3dDeconvolve with 3dREMLfit to correct for spurious significance resulting from serial correlation of noise. As with the AFNI powerpoint on advanced 3dDeconvolve, use of 3dREMLfit dramatically drops the individual subject voxel-wise significance level, but doesn't seem to impair the significance of group-wise maps.
My question is, what is a good way to describe this correction in methods. Has anyone used 3dREMLfit in an actual paper? I could use a ref or two to see how this has been described for clarity. The issue arises because I would be writing about a patient population in a clinical journal, where reviewers and readership may appreciate a simplified translation of ARMA(1,1) and what that means.
Also, a validation paper of this method itself would be helpful to cite. There appears to be no reference in the -help on 3dREMLfit.
Jim Bjork