Glutamate (Glu) and γ-amino butyric acid (GABA) are major neurotransmitters in human brain. These peaks are overlapped on the conventional ¹H spectra due to strong coupling with small chemical shift difference and J_HH coupling. This may lead to difficulty of accurate absolute quantitation of Glu and GABA. Constant time (CT) two dimensional methods have a feature of good peak resolution through ¹H decoupling along F₁. We have reported in vivo detection of these metabolite peaks of Glu C4H at 2.35ppm, GABA C2H at 2.28ppm and glutamine (Gln) C4H at 2.44ppm in human brain using two kinds of 2D localized CT methods, CT-COSY (1) and CT-PRESS (2). Since long constant time delay (T_ct) is required in these methods, T₂ correction is needed for absolute quantitation. To overcome this, we have proposed 2D FT with shared time domain (TD) data and demonstrated quantitation of Glu (3). In this work, we will expand this method into measurement of concentration of Glu and GABA and demonstrate absolute quantitation in phantom experiments.

In CT-PRESS sequence, water suppression and outer volume suppression are followed by a module for PRESS localization; 90° slice pulse (y-direction) – 1/2*TE1 – 180° -slice pulse (x-direction) – 1/2*(TE1+TE2)+Δt₁/2 – 180° slice pulse (z-direction) – {data acquisition}. To meet the constant time condition, a suitable amount of zeroes were filled in front of the acquisition data (2). Figure 1 shows a schematic of our proposed quantitation method. After N₁ increments by Δt₁ along t₁, 2D TD data defined by N₁ × n₂ matrices are accumulated where n₂ is number of sampling points in FID. When a part of the total TD data defined by n₁ × n₂ shown as a dotted area of TD1 in Fig. 1 is extracted, reconstruction of the TD1 data generates a CT-PRESS spectrum weighted by exp(-T_ct/T₂) where the value of T_ct is TE1+TE2+(n_1start+n₁/2)Δt₁. Since T_ct can be varied by shifting n_1start, series of ¹H decoupled spectra weighted with multiple T_ct's can be reconstructed from series of the partial TD data. Figure 1 shows an example of two sets of TD data of TD1 and TD2. Next, peak volumes with multiple T_ct's are calculated by curve-fitting of these spectra (4) using a linear combination model with a basis set. In this work, we used three bases of Glu, GABA and Gln. Calculated peak volumes can be expressed as an exponential decay and T₂ correction can be done for measurement of concentrations of metabolites.

All experiments were performed using a 4.7 T whole-body MR system (INOVA, Agilent). First, we measured each CT-PRESS spectrum of 50-mM Glu, GABA and Gln for the basis set. For demonstration of quantitation, we measured a phantom containing a brain metabolite mixture of 10.0 mM NAA, 8.0 mM Cr, 9.0 mM Glu, 2.6 mM GABA and 2.9 mM Gln. A 100-mL bottle containing this solution was placed in a water bath and CT-PRESS signals were acquired inside a voxel within that bottle. After a CT-PRESS spectrum was reconstructed from the partial TD data, the area from 2.2ppm to 2.6ppm was extracted for curve fitting (Fig. 2) and absolute quantitation was done. CT-PRESS signals were acquired in an 8ml voxel with a measurement time of 33 min. In all measurements, TE1 and TE2 were 17.5 ms. Spectral widths along F₁ and F₂ were 1 kHz and 2 kHz, respectively. N₁ and n₁ were 200 and 150, respectively. Relaxation delay was 5 s.

Figure 2 shows a phased CT-PRESS spectrum with T_ct = 109 ms of the mixture phantom with views from F₁ and F₂ axes. By reconstruction with a window function of resolution enhancement for shifted echoes (2), Glu C4H, GABA C2H and Gln C4H are resolved. First, peak volumes of these metabolites were calculated by curve fitting of CT-PRESS spectra with T_ct’s of 109, 119, 129, 139, 149 and 159 ms. Then, calculated peak volumes were curve fitted by an exponential decay model for T₂ correction (Fig. 3). Finally, the values of 9.5 mM, 2.5 mM and 3.7 mM were obtained for Glu, GABA and Gln, respectively. While these values of Glu and GABA are close to each concentration of mixture phantom, the value of Gln is 28% bigger than the concentration. This error may be caused by the NAA multiples around 2.5 ppm.Our proposed method is useful to absolute quantitation of Glu and GABA. Since these peaks are resolved even in the human brain spectra, this method will also be applied in human studies with the internal water reference method.