Multi-echo AxCaliberSMT: Mapping effective axon radius from restricted diffusion and radius-dependent intra-axonal T2
Kwok-Shing Chan, Yohan Jun, Aneri Bhatt, Berkin Bilgic, Susie Y. Huang, Hong-Hsi Lee
Abstract
Diffusion-based axon radius r mapping preferentially detects larger axons since restricted-diffusion contrast scales with r4, limiting sensitivity to thin axons dominating human white matter (WM). Surface relaxation provides complementary sensitivity through an intra-axonal R2 proportional to 1/r. We combined these mechanisms to broaden the dynamic range of radius estimations. Multi-echo AxCaliberSMT (ME-AxCaliberSMT) models spherical-mean diffusion-relaxation signals using cylinder-restricted diffusion and histologically calibrated surface relaxation. Simulations compared its accuracy and robustness with AxCaliberSMT and a diffusion-filtered R2 model, which infer radius solely from restricted diffusion and intra-axonal R2, respectively. Five healthy adults were scanned on Connectome 2.0 (C2; 500 mT/m), and four also on Cima.X (200 mT/m), using multiple b-values at two diffusion times, each paired with a different echo time. Diffusion and relaxation contrasts were comparable at a protocol-dependent transition radius (C2: 1.42 um; Cima.X: 1.85 um). Below this threshold, relaxation dominated, yielding effective r estimates approximately <r2>/<r> in most WM. At SNR 40, ME-AxCaliberSMT achieved normalised bias 6%, whereas AxCaliberSMT overestimated small radii. ME-AxCaliberSMT was less sensitive to tested calibration errors than the R2 model. In vivo, ME-AxCaliberSMT bundle radii on C2 were 0.30-0.51 um (AxCaliberSMT: 1.03-1.38 um) and followed the histological pattern along the corpus callosum. ME-AxCaliberSMT estimates correlated between scanners (mean Pearson correlation 0.73), unlike AxCaliberSMT (-0.23), although Cima.X values were systematically higher. Combining diffusion and relaxation extends axon radius mapping from large to small axons. Translation to a commercial high-gradient scanner reproduced bundle-wise patterns, although absolute estimates remained protocol-dependent.
Create a lesson
Related papers
Towards consensus on practical probabilistic evaluation metrics for clinical decision making in radiotherapy
Jenneke I. de Jong, Steven J. M. Habraken, Erik W. Korevaar et al.
Unmyelinated axons alter the biological interpretation of MRI biomarkers of demyelination
Ali Abdollahzadeh, Ricardo Coronado-Leija, Elizabeth Chasen et al.
Impact-Based Analysis Method (IBAM) to evaluate the primary stability of the pedicle screw: A proof of concept
Julie Gabriel, Yvan Petit, Giuseppe Rosi et al.
Geometry-bridge flow matching for blind rigid-motion correction in head cone-beam CT
Sungho Yun, Seungryong Cho
The relationship between junctional zone, smooth muscle tone and intra-uterine pressure
Nicolas R. Chevalier, Emilie Huon, Julie Poujol et al.
A Soft Sponge-Pillar Electrode Platform for Motion-Tolerant Ambulatory EEG and Sleep Monitoring
Chansoo Kim, Weilun Li, Ryan Andersen et al.