I was looking at different quality assurance pipelines recently and found differences in the way framewise displacement (Powers,2012) was calculated.
In Poldrack's QA-scripts: https://github.com/poldrack/fmriqa/blob/master/compute_fd.py/https://github.com/nipy/nipype/blob/98a09d5762871bbd91e3a9af3c68ce8117004ae3/nipype/algorithms/confounds.py#L213 it is assumed that rotational displacement is in the unit of radian (disp[:,0:3]=N.pi*headradius*2*(disp[:,0:3]/(2*N.pi)) OR diff[:, :3] *= self.inputs.radius. In another publicly available script https://github.com/juhuntenburg/mriqc/commit/67e1eb044cd0d709e0b26a2b1461a9916a00159f the unit of the rotational displacement is assumed to be degrees.
I wondered whether it is documented for MCFLIRT and AFNI motion correction tool if the rotational parameters are in radians or in degrees? I couldn't find a proper source.
Thank you for your help,