Navigator readouts with optimized processing improve spinal cord and gray matter segmentability in multi-echo gradient-echo imaging.
Büeler Silvan S, Beghini Laura L, Kündig Christian W CW, Liechti Martina D MD et al.
Multi-echo gradient-echo (ME-GRE) acquisitions of the spinal cord (SC) are highly susceptible to breathing-induced magnetic field fluctuations arising from the proximity of the lungs, often resulting in ghosting artifacts and signal loss. A navigator readout can be employed to monitor and correct for these field variations; however, conventional navigator processing methods frequently fail in the SC due to its small size and distinct anatomy. Here, we evaluate the impact of a recently introduced, SC-specific navigator processing approach on SC and gray matter (GM) segmentability. ME-GRE images were acquired at 3T during free breathing in ten healthy volunteers, covering the cervical, thoracic, and lumbosacral spinal cord. Data were collected both with and without a centerline navigator readout. Acquisitions without navigators included five imaging echoes, whereas those with navigators comprised four imaging echoes followed by an additional readout of the k-space centerline (the "navigator echo"). The navigator readout was processed using a recently introduced method optimized for the SC. The effect of the optimized navigator correction was evaluated by contrast-to-noise ratio (CNR), qualitative ratings of SC and GM segmentability, and the performance of automatic segmentation algorithms. Compared with acquisitions without navigators, optimized navigator correction significantly improved quantitative image quality metrics, particularly GM-white matter CNR at the C7-T1 and T7-T8 vertebral levels and across the lumbosacral cord. Regarding SC and GM segmentability, raters preferred images with navigators over uncorrected images at nearly all spinal cord levels, with the strongest effect observed for GM segmentability in the lumbosacral cord. Although overall automatic segmentation performance was comparable between acquisition types, automatic GM segmentation failed at the lumbosacral cord in the presence of severe artifacts in acquisitions without navigators, while remaining robust in the corresponding navigator-corrected images. In conclusion, navigator processing optimized for SC imaging improves image quality and segmentability in ME-GRE acquisitions, making this technique attractive for both clinical and research applications.