31P MRSI Coil Combination Using 23Na Sensitivity Information

TL;DR

This paper introduces a new method to improve the signal quality in 31P MRSI by using 23Na sensitivity data for coil combination.

Contribution

A novel coil combination technique for 31P MRSI using high-SNR 23Na sensitivity maps instead of low-SNR self-weighting.

Findings

Using 23Na sensitivity maps results in negligible SNR loss (max 5%) in 31P MRSI.

The 23Na-based method avoids SNR overestimation bias seen in self-weighting approaches.

In vivo experiments show improved SNR with the proposed method across three acquisitions.

Abstract

To optimally combine receive channels for 31P MRSI when the SNR is relatively low using sensitivity maps obtained from 23Na signals.

This study utilized a dedicated X‐nuclei 15‐channel head coil array capable of imaging both 23Na and 31P. After acquiring the relatively low‐SNR multi‐channel 31P MRSI signals, also the high‐SNR 23Na sensitivity information was acquired. The low‐SNR 31P MRSI signals were combined using the high‐SNR 23Na sensitivity information, instead of the traditional self‐weighting approach where 31P sensitivity is derived from the low‐SNR 31P FID signals. The proposed method was evaluated by electromagnetic simulations, numerical Monte Carlo studies, and in vivo 31P MRSI on two healthy volunteers undergoing three independent acquisitions in total.

The electromagnetic simulations indicate a negligible SNR loss of maximum 5% by using 23Na sensitivities for 31P coil combination. The Monte Carlo synthetic coil combination shows that the higher SNR of the 23Na sensitivity information benefits the SNR of the combined 31P spectra. In addition, the self‐weighted method is prone to introducing combination bias (SNR overestimation), while the 23Na‐based method does not. The in vivo MR experiments demonstrate an increase in SNR when using the 23Na‐based combination, as verified across three independently acquired datasets.

31P MRSI multi‐channel signal combination using 23Na sensitivities acquired with the same receiver array provides better performance when compared to self‐weighting.

Figures