Field, coil, and echo-time influence on sensitivity and reproducibility of brain proton MR spectroscopy.

BACKGROUND AND PURPOSE Clinical MR imaging scanners now offer many choices of hardware configurations that were not available in the first 25 years of their existence. Our goal was to assess the influence of coil technology, magnetic field strength, and echo time (TE) on the sensitivity, reflected by the signal intensity-to-noise-ratio (SNR) and reproducibility of proton MR spectroscopy (1H-MR spectroscopy). MATERIAL AND METHODS The SNR, the intersubject reproducibility, and the intrasubject reproducibility of N-acetylaspartate (NAA), creatine (Cr), and choline (Cho) levels were compared at the common TEs of 30, 144, and 288 ms, by using 1H-MR spectroscopy in 6 volunteers at (1) 3T with a single-element quadrature (SEQ); (2) 1.5T with SEQ; and (3) 1.5T with a 12-channel phased-array (PA) head coil. RESULTS In terms of sensitivity, the best SNR for all metabolites was obtained at the shortest TE (30 ms). It was comparable between the 3 and 1.5T with the PA, but approximately 35% better than the 1.5T with SEQ. This SNR difference declined <25% at TE of 144 ms and to equity among all imagers at TE of 288 ms. Reproducibility, reflected in the coefficient of variation (CV), was best for NAA at TE of 288 ms, 15%-50% better than at TE of 30 ms in either gray (GM) or white matter (WM). The CV for Cr was best, at TE of 288 ms for GM, but its WM results were independent of TE. Metabolite level reproducibility did not depend on coil technology or magnetic field strength. CONCLUSIONS For the same coil type, the SNR of all major metabolites was approximately 35% better at 3T than at 1.5T. This advantage, however, was offset at 1.5T with a PA coil, making it a cost-effective upgrade for existing scanners. Surprisingly and counterintuitively, despite the lowest SNR, the best reproducibility was obtained at the longest TE (288 ms), regardless of field or coil.