Axon Diameter Mapping in the Living Human Brain with Ultra-High-Gradient Diffusion MRI at 500 mT/m Gradient Strength

Pushing the resolution limit of axon diameter mapping in the living human brain requires higher gradient strength than is currently available on most clinical MRI systems. The noninvasive quantification of axon diameter not only enables the exploration of axonal damage in a wide range of neurological disorders but also provides fundamental insights into axonal organization and conduction patterns of white matter tracts. The goal of this study was to evaluate the sensitivity of axon diameter mapping to small diameter axons using the next-generation Connectome MRI scanner (Connectome 2.0), which features a maximum gradient strength of 500 mT/m and slew rate of 600 T/m/s, compared to the original Connectome 1.0 scanner with 300 mT/m gradient strength. We applied the AxCaliber-SMT model to diffusion MRI data from 40 healthy adults, comprising 20 participants scanned on Connectome 1.0 before it was decommissioned and a separate cohort of 20 age- and sex-matched participants scanned on Connectome 2.0. Our findings are based on group-level comparisons between these two cohorts. The theoretical minimum detectable axon diameter was 2.5 μm on Connectome 2.0 and 3.6 μm on Connectome 1.0. The MR-estimated axon diameter in the corticospinal tract on the Connectome 2.0 scanner was 2.66 ± 0.54 μm, significantly lower than 3.35 ± 1.00 μm on Connectome 1.0 (Welch's t-test: p = 0.0110). Comparison was performed across independently acquired datasets from age- and sex-matched individuals on different scanners; therefore, the observed group differences should be interpreted as strong and supportive, though not strictly causal, evidence of the proposed scanner capability. Furthermore, we examined 7 healthy adults for scan-rescan repeatability and demonstrated that the voxel-wise mean absolute difference in axon diameter estimates between scan and rescan decreased to 0.29 μm on the Connectome 2.0 (vs. 0.65 μm on the Connectome 1.0), indicating improved repeatability of the axon diameter estimates. These improvements are enabled not only by the higher gradient strength of Connectome 2.0, but also by the associated reduction in echo time and increase in SNR, which together enhance sensitivity to restricted diffusion and improve parameter reliability. Our findings highlight the importance of stronger and faster gradients for accurate and robust mapping of the axonal microstructure in the human brain.