Dynamic Shimming in the Cervical Spinal Cord for Multi-Echo Gradient-Echo Imaging at 3 T

TL;DR

This paper introduces a dynamic shimming technique for spinal cord MRI at 3 T, combining in-plane and through-plane adjustments with real-time respiration-based corrections to enhance image quality.

Contribution

It develops a novel real-time dynamic shimming method that improves magnetic field homogeneity in spinal cord imaging by incorporating in-plane gradients and respiration adjustments.

Findings

Improved signal homogeneity with z-shimming.

Further enhancement with real-time respiration compensation.

Potential for better image quality in clinical settings.

Abstract

Obtaining high quality images of the spinal cord with MRI is difficult, partly due to the fact that the spinal cord is surrounded by a number of structures that have differing magnetic susceptibility. This causes inhomogeneities in the magnetic field, which in turn lead to image artifacts. In order to address this issue, linear compensation gradients can be employed. The latter can be generated using an MRI scanner's first order gradient coils and adjusted on a per-slice basis, in order to correct for through-plane ("z") magnetic field gradients. This approach is referred to as z-shimming. The aim of this study is two-fold. The first aim was to replicate aspects of a previous study wherein z-shimming was found to improve image quality in T2*-weighted echo-planar imaging. Our second aim was to improve upon the z-shimming approach by including in-plane compensation gradients and adjusting the compensation gradients during the image acquisition process so that they take into account respiration-induced magnetic field variations. We refer to this novel approach as realtime dynamic shimming. Measurements performed in a group of 12 healthy volunteers at 3 T show improved signal homogeneity along the spinal cord when using z-shimming. Signal homogeneity may be further improved by including realtime compensation for respiration-induced field gradients and by also doing this for gradients along the in-plane axes.