Navigation and service

Optimization of high-channel count, switch matrices for multinuclear, high-field MRI

Jörg Felder, Chang-Hoon Choi, Yunkyoung Ko and N. Jon Shah

19th August 2020

Magnetic resonance imaging (MRI) at a field strength of between 0.5 and 3 T is now part of standard clinical practice and enables doctors to image the human body non-invasively and with a high level of detail. However, MRI is an evolving technology and a great deal of current research is looking at ways to improve the quality of the attained images by increasing the magnetic field strength to 7 T and beyond (high and ultra-high field strength). Another significant area of research and development is related to the acquisition of images using the signal from both proton and non-proton nuclei (multinuclear MRI) in order to obtain information relating to metabolic processes in the body.

In hospital situations, MRI is often used on severely ill patients with reduced tolerance to endure long examinations, so that obtaining the maximum amount of information in the shortest time possible is paramount. In modern MRI systems with field strengths below 3 T, this is partly achieved using parallel imaging techniques with high-channel count, phased array coils. Phased array coils use a number of smaller coils which are fed into separate receiver chains via connectors located on the patient table. Information from each coil is combined digitally to form a single image which covers a large field of view.

For several practical, economic and safety reasons, it is better to use a switch matrix to only route a subset of the connectors to a small number of signal receivers. However, the use of switch matrices is not common practice in the new state-of-the-art ultra-high field MRI systems, and furthermore, research has so far focussed on single frequency designs that can be used solely to route the signal of the predominantly employed proton nucleus. With the increasing interest in developing multinuclear, multichannel coil arrays; switch matrices capable of handling multiple frequencies are desirable.

In response to this challenge, this work presents the use of metaheuristic approaches to optimise the circuit design of these matrices and, for a matrix with 128 inputs and 64 outputs. A solution is proposed that displays a worst-case insertion loss of 3.8 dB. This means that the quality of images acquired with the presented approach are not detrimentally affected by the comparatively small decrease in SNR caused by receive signal routing.

By demonstrating that circuit optimization of a large, multi-nuclear switch matrix is feasible, it is anticipated that, following experimental validation, the approach could support clinical practice in the future by improving the quality of MRI images from ultra-high field (UHF) systems while avoiding an increase in scanning time.

Original publication:

Optimization of high-channel count, switch matrices for multinuclear, high-field MRI