Optimized navigator-based correction of breathing-induced B₀field fluctuations in multi-echo gradient-echo imaging of the spinal cord

Abstract

Purpose: Multi-echo gradient-echo (ME-GRE) imaging in the spinal cord is susceptible to breathing-induced B0field fluctuations due to the proximity of the lungs, leading to ghosting artifacts. A navigator readout can be used to monitor the fluctuations; however, standard navigator processing often fails in the spinal cord. Here, we introduce navigator processing tailored specifically for spinal cord imaging. Methods: ME-GRE data covering all spinal cord regions were acquired in six healthy volunteers during free breathing at 3T. Centerline navigator readouts and respiratory belt recordings were collected during the acquisitions. The navigator processing included a Fast Fourier Transform and subsequent interval selection targeting the spinal cord, as well as SNR-weighted averaging over samples and coils on the complex data. Furthermore, a phase unwrapping algorithm making use of the belt recordings was developed. Imaging data were corrected by phase demodulation before image reconstruction. Results: B0field fluctuations and ghosting artifacts were largest in the lower cervical and upper lumbosacral cord (~5 Hz std), close to the edges of the lungs. Image reconstruction based on optimized navigator correction improved visual image quality and quantitative metrics (SNR, CNR, ghosting) in all regions of the spinal cord. The improvement was largest in regions with large field fluctuations (group-averaged increase in SNR/CNR of up to 29%/37% in single-echo images). Conclusions: Optimized navigator-based correction can reduce ghosting artifacts and increase SNR/CNR in anatomical ME-GRE of the spinal cord. The enhanced image quality and ease of implementation across sites makes the technique attractive for clinical and scientific applications.