Optimization of metal artefact reduction (MAR) sequences for MRI of total hip prostheses

To describe the relative contribution of matrix size and bandwidth to artefact reduction in order to define optimal sequence parameters for metal artefact reduction (MAR) sequences for MRI of total hip prostheses.

A phantom was created using a Charnley total hip replacement. Mid-coronal T1-weighted (echo time 12ms, repetition time 400ms) images through the prosthesis were acquired with increasing bandwidths (150, 300, 454, 592, and 781Hz/pixel) and increasing matrixes of 128, 256, 384, 512, 640, and 768 pixels square. Signal loss from the prosthesis and susceptibility artefact was segmented using an automated tool.

Over 90% of the achievable reduction in artefacts was obtained with matrixes of 256x256 or greater and a receiver bandwidth of approximately 400Hz/pixel or greater. Thereafter increasing the receiver bandwidth or matrix had little impact on reducing susceptibility artefacts. Increasing the bandwidth produced a relative fall in the signal-to-noise ratio (SNR) of between 49 and 56% for a given matrix, but, in practice, the image quality was still satisfactory even with the highest bandwidth and largest matrix sizes. The acquisition time increased linearly with increasing matrix parameters.

Over 90% of the achievable metal artefact reduction can be realized with mid-range matrices and receiver bandwidths on a clinical 1.5T system. The loss of SNR from increasing receiver bandwidth, is preferable to long acquisition times, and therefore, should be the main tool for reducing metal artefact.