Reducing Inter-Individual Differences in Task fMRI Preprocessing with OGRE (One-Step General Registration and Extraction) Preprocessing

Volumetric analysis methods continue to enjoy great popularity in the analysis of task-related functional MRI (fMRI) data. Among these methods, omnipresent throughout the field is the software tool FEAT, part of the FSL package (FMRIB, Oxford, UK). However, FEAT preprocessing involves multi-step interpolation (multiple sequential transformations), which at typical fMRI field strengths (i.e. 3T) has not been systematically compared with with alternative methods, in part due to lack of software tools compatible with FSL FEAT analysis. Here we developed the One-step General Registration and Extraction (OGRE) pipeline to combine FreeSurfer brain extraction, FSL FNIRT registration, one-step interpolation of preprocessing transformations, and output for FSL FEAT analysis. We compared OGRE with two alternative preprocessing methods (fMRIPrep and FSL) in a dataset wherein adult human volunteers (N = 53) performed a precision drawing task during fMRI scanning. OGRE preprocessing led to lower inter-subject variability than FSL (p = 7.3 × 10^{− 9}) or fMRIPrep (p = 0.036); OGRE also led to the strongest detection of task-related magnitude in primary motor cortex contralateral to movement (OGRE > FSL p = 4.2 × 10^{− 4}, other method pairs not significant). This pattern suggests that one-step interpolation (used by OGRE and fMRIPrep) reduces inter-individual variability during moderate-resolution (2 mm³) volumetric analysis of task fMRI data.