Sports concussions account for up to 30% of all traumatic brain injuries, yet their subacute pathophysiology remains poorly understood. Earlier QSM studies focused on large deep brain structures such as the putamen, caudate nucleus, and globus pallidus, and they produced conflicting results on iron accumulation. This work sits alongside studies hunting for injury biomarkers in the deep brain nuclei, but it stands out by being the first to divide the hippocampus into subfields and to apply sign-based voxel thresholding so that opposing magnetic signals are not mixed together.
Participants underwent MRI on a 3-tesla Siemens Vida Fit scanner with a QSM sequence. Susceptibility maps were reconstructed with the QSMxT pipeline, then voxels were split into positive (iron) and negative (myelin, calcium, proteins). For 16 basal ganglia segmentations, the CIT168 atlas was used, and for 10 hippocampal subfields, automatic FreeSurfer segmentation based on an ultra-high-resolution ex vivo atlas. Statistics were run with correction for multiple comparisons using the Benjamini-Hochberg method.
No significant differences between the groups appeared in positive susceptibility in any of the 16 deep brain nuclei or any of the 10 hippocampal subfields. In the CA4 subfield, negative susceptibility in injured participants was significantly less negative (mean -0.007 ppm versus -0.008 ppm in controls, q=0.04). Correlations with age showed the expected increase in positive susceptibility in the putamen (r=0.64), caudate nucleus (r=0.50), red nucleus (r=0.66), PBP (r=0.40), and ventral pallidum (r=0.35). Negative susceptibility increased with age in the fimbria of the hippocampus (r=0.42), caudate nucleus, and extended amygdala, but decreased in the ventral pallidum. No links were found with symptom severity (BIST) or time since injury.
The absence of differences in positive susceptibility suggests that acute iron accumulation in the deep brain nuclei is not a universal response to mild injury, unlike in the cortex, where changes had been found earlier. The drop in negative susceptibility in CA4 fits the diffuse axonal injury model and may reflect demyelination or loss of vulnerable neurons. The work shows that separating signals by voxel sign makes it possible to see effects that get lost when an entire region is averaged.
The topic could move toward longitudinal studies with repeated scans to track how changes unfold from the acute phase into the chronic one. Using multi-echo QSM sequences would make it possible to separate signal sources within a single voxel and detect subtler changes. There are also plans to include female athletes to assess sex differences in the response to injury.
The results matter for sports medicine and neurology: they clarify which brain structures are most vulnerable in the first two weeks after a concussion and could help in developing more sensitive diagnostic methods.
Run a longitudinal follow-up of the same participants at 3, 6, and 12 months. Repeat the analysis with a multi-echo QSM sequence to separate signal sources within a voxel.
The work tackles an unresolved question: why do some people develop persistent cognitive problems after mild injury while others do not? The lack of correlation between symptom severity and susceptibility changes suggests that subjective scales may not capture objective pathophysiological processes, and that more sensitive biomarkers are needed.
The hippocampus is named after the seahorse because of its shape, and the CA4 subfield sits in its "tail" and contains cells that are among the first to die in epilepsy and injury.
Not published on arXiv. Here are its page, the PDF and the formalization: you can check it yourself, not only read our account.