The test-retest reliability of two-dimensional (2D) correlated spectroscopy (COSY) was studied on a whole-body 1.5T MRI scanner. Single-voxel localized 2D proton spectra were recorded in vitro as well as in vivo using a recently implemented localized chemical shift correlated spectroscopic (L-COSY) sequence. A total of 40 in vitro and 40 human brain (10 volunteers, four times each) 2D L-COSY spectra were recorded. The coefficients of variation (CVs) of selected brain metabolites (raw volume integrals) recorded in 10 healthy volunteers were less than 9% for creatine, choline, and N-acetyl aspartate, and less than 17% for myo-inositol, glutamine/glutamate, aspartate, and threonine/lactate. The 2D metabolite ratios and the raw volume integrals of 2D diagonal and cross peaks in healthy human brain were very well reproduced. The intraclass correlation coefficients were greater than 0.4 (P < 0.05) for the major metabolites, indicating that the 2D peak volumes were stable enough within individuals to detect reliable differences between normal subjects. In the past decade, MR spectroscopy (MRS) has become a reliable clinical tool for the diagnosis of selected diseases (1-7). An important feature of any diagnostic tool is its reproducibility. Many studies on the reproducibility of one-dimensional (1D) 1 H MRS have reported the ratios and absolute concentrations of the metabolites (8 -13). Although these studies have shown small errors (3-6%) in the measurement of metabolites in vitro (8,9), the in vivo studies showed higher coefficients of variation (7-26%) in a maximum of only five metabolites (9 -13). Severe overlap of brain metabolites is a major hindrance in 1D MRS.Localized and nonlocalized versions of 2D MRS were reported a decade ago (14 -18); however, only recently have 2D proton spectra been recorded in healthy human subjects and patients with brain tumors (19 -25). Compared to 1D MRS, 2D MR spectra allow less ambiguous assignment of several metabolite resonances in human brain and prostate (25)(26)(27). In a previous study (25), the resonances caused by glutamine/glutamate (Glx) were clearly separated from the dominant singlet resulting from N-acetyl aspartate (NAA) and myo-inositol (mI) in the human brain. Even though identification of "free" aspartate (Asp) was not possible with 1D MRS, the peaks were separated from NAA in 2D MRS (25). 2D cross peaks caused by phosphoethanolamine and ethanolamine (PE), phosphoryl choline (PCh), threonine and lactate (Thr/Lac), and ␥-aminobutyric acid (GABA) were also identified in the 2D L-COSY spectra of the brain. However, to date there has been no report on the reproducibility of 2D MRS. The goal of this work was to investigate the reproducibility of 2D peak volumes and 2D metabolite ratios recorded in phantom solutions and healthy brain using the L-COSY sequence (25).
MATERIALS AND METHODS
A 1.5 T GE Horizon (5.8) MRI/MRS scanner (GE MedicalSystems, Waukesha, WI) with echo-speed gradients was used with a body coil for transmission and a 3-inch surface coil for reception. ...