Multiscale structural MRI differences among collegiate American football players, cross-country runners, and non-athletes

Abstract

Structural differences observed in collision-sport athletes may reflect repetitive head-impact exposure, broader effects of athletic training, or pre-existing individual variation, making inclusion of a non-collision athlete group important for interpretation of any differences seen. Here, structural T1-weighted MRI data from 17 collegiate American football players, 16 cross-country runners, and 9 non-athletes were analyzed using FSL voxel-based morphometry (VBM), FIRST subcortical volumetry and vertex-wise shape analysis, and FAST whole-brain tissue segmentation. Spatial inference used permutation testing with threshold-free cluster enhancement and family-wise error correction, while scalar analyses used multiplicity-corrected procedures. FAST-derived absolute tissue volumes were also examined in a sensitivity analysis adjusting for FreeSurfer estimated total intracranial volume (eTIV). VBM showed recurrent brainstem-centered differences, with non-athletes showing greater local gray matter measures than both athlete groups, whereas football players showed greater brainstem measures than runners. FIRST vertex-wise analysis identified corrected differences involving the brainstem and selected subcortical structures, while no total FIRST structure volume survived false discovery rate correction. Unadjusted FAST analysis identified a CSF group effect, F(2, 24.38) = 11.46, p < .001, q = .001. After eTIV adjustment, the CSF group effect remained, F(2, 38) = 14.30, q < .001, and a white matter group effect was also identified, F(2, 38) = 9.59, q < .001; gray matter and total segmented tissue did not show corrected adjusted effects. These findings indicate that structural variation depends on anatomical scale, comparison group, and head-size adjustment and cannot be attributed uniquely to collision-sport exposure.