Effect of Carr-Purcell refocusing pulse trains on transverse relaxation times of metabolites in rat brain at 9.4 T

Abstract

Purpose

To investigate the effect of Carr-Purcell (CP) pulse trains on transverse relaxation times, T₂, of tissue water and metabolites (both non-coupled and J-coupled spins) in the rat brain at 9.4 T using LASER, CP-LASER and T_2ρ-LASER sequences.

Methods

Proton NMR spectra were measured in rat brain in vivo at 9.4 T. Spectra were acquired at multiple echo times ranging from 18 to 402 ms. All spectra were analyzed using LCModel with simulated basis sets. Signals of metabolites as a function of echo time were fitted using a mono-exponential function to determine their T₂ relaxation times.

Results

Measured T₂s for tissue water and all metabolites were significantly longer with CP-LASER and T_2ρ-LASER compared to LASER. The T₂ increased by a factor of ~1.3 for non-coupled and weakly coupled spins (e.g., N-acetylaspartate and total creatine) and by a factor of ~2 (e.g., glutamine and taurine) to ~4 (e.g., glutamate and myo-inositol) for strongly coupled spins.

Conclusion

Application of a CP pulse train results in a larger increase in T₂ relaxation times for strongly coupled spins than for non-coupled (singlet) and weakly coupled spins. This needs to be taken into account when correcting for T₂ relaxation in CP-like sequences such as LASER.

INTRODUCTION

Localization by adiabatic selective refocusing (LASER) (1) and semi-LASER (2) sequences are widely used for in vivo single-voxel ¹H NMR spectroscopy (MRS) studies in both humans and animals (3-9). These sequences offer excellent localization performance and greatly reduced chemical-shift displacement error, as well as diminished sensitivity to B₁ inhomogeneity (i.e., adiabaticity).

The succession of multiple adiabatic full-passage (AFP) pulses in LASER and semi-LASER constitutes a Carr-Purcell (CP) train (10) which is known to minimize J-evolution for J-coupled metabolites (11). The time evolution due to J-modulation becomes negligible when ${\tau }_{\mathrm{cp}}\cdot \sqrt{{\delta }^2+J^2}\ll 1$ where τ_cp is an interpulse delay (in s), J is the coupling constant (in Hz), and δ is the chemical shift difference (in Hz) between the coupled spins. The preservation of the spectral pattern for several J-coupled metabolites e.g., glutamate, glutamine and myo-inositol (mIns) has been demonstrated successfully at relatively long echo times (T_E) in the human brain using CP-like PRESS sequences at 1.5 T (12) and 2 T (13). CPRESS (T_E = 45 ms) was also shown to improve detection of mIns in patients with cognitive impairment compared to the standard PRESS sequence at 3 T (14).

In addition to minimizing J-modulation, CP pulse trains also refocus the effects of diffusion and chemical exchange (15), resulting in longer apparent transverse relaxation times (T₂) than with standard sequences such as STEAM or PRESS. Accordingly, an increase in the apparent T₂ of tissue water in the human brain was reported using CP-LASER at 4 T and 7 T compared to PRESS (5) and LASER (16). Similarly, an increase in apparent T₂s was reported for non-coupled spins of N-acetylaspartate (NAA) and total creatine (tCr) with CP-LASER compared to PRESS (5). The term, apparent T₂, refers to the observed decay constant of the transverse magnetization, irrespective of the physical processes that contribute to the signal decay. Henceforth, terms, apparent T₂, T₂, and transverse relaxation time, will be used interchangeably.

In spite of the increasing popularity of CP-like sequences such as CP-PRESS, LASER and semi-LASER, however, there have been very few studies of the effect of CP pulse trains on T₂s of brain metabolites other than singlets. One study reported T₂s of mIns and the multiplet of NAA in human brain with CP-PRESS at 1.5 T (12). In that study, the authors noted that T₂ of mIns with CP-PRESS was somewhat longer than previously published with PRESS or STEAM. However, no precise quantitative comparison was made between PRESS and CP-PRESS, and T₂s for J-coupled metabolites were limited to mIns and the multiplet of NAA. Therefore, the quantitative effect of CP trains on T₂s of other coupled metabolites, such as glutamate, glutamine or taurine, remains to be determined.

The aim of the present study was to investigate the effect of CP pulse trains on the T₂s of metabolites (both non-coupled and J-coupled spins) in the rat brain at 9.4 T using LASER, CP-LASER and T_2ρ-LASER sequences.

METHODS

Male Sprague-Dawley rats (n = 6) were studied using a 9.4 T horizontal-bore magnet (Magnex Scientific, Oxford, UK) interfaced to an Agilent DirectDrive console (Agilent Technologies, CA, USA). The magnet was equipped with a gradient insert capable of reaching 400 mT/m in 300 μs (Resonance Research, Inc., MA, USA). The animals were anesthetized with 1.5% isoflurane in a mixture of 50% O₂:50% N₂O and placed in a cradle with the head fixed with bite-bar and ear rods. The body temperature was monitored throughout the study with a rectal thermal sensor and maintained at 37°C using a hot water circulation. All studies were approved by the Institutional Animal Care and Use Committee at the University of Minnesota.

Localized in vivo ¹H NMR spectra from the brain were measured using LASER, CP-LASER and T_2ρ-LASER sequences (Figure 1). A volume-of-interest (VOI) of 62.5 μL (5 × 2.5 × 5 mm³) was positioned on the midline 2 mm posterior to bregma and 3 mm ventral using transverse and sagittal RARE (17) images (repetition time, T_R = 4 s, T_E = 20 ms, echo train length = 8, field-of-view = 3 × 3 cm², matrix size = 256 × 128, slice thickness = 1 mm, 10 slices) and included tissue from cortex, hippocampus, and striatum. For all NMR measurements, a home-built quadrature 400 MHz ¹H surface RF coil (two loops, 14 mm diameter each) was used. B₀ shimming in the defined VOI was performed using an adiabatic version of FAST(EST)MAP (18) which resulted in a water linewidth of 13 ± 1 Hz.

Figure 1.

Diagrams of A) LASER, B) CP-LASER and C) T_2ρ-LASER sequences used for MRS acquisition. In LASER, the AFP pulses were applied in the presence of gradients for 3D localization (represented by the gray box). In CP- and T_2ρ- LASER sequences, a CP train of AFP pulses (identical adiabatic pulses as utilized in LASER) was inserted prior to the LASER localization and was phase cycled according to the MLEV scheme. The τ_cp delay was 3 ms in CP-LASER and 1.5 ms in T_2ρ-LASER (equivalent to the duration of one AFP pulse). No spoiler gradients were applied during the CP segment.

The LASER sequence consisted of a 4 ms nonselective adiabatic half-passage (AHP) pulse followed by three pairs of gradient selective AFP pulses of the HSn family (1.5 ms duration, n = 1, R = 25, (19)) for 3D-localization (Figure 1A). T_E was defined as the total time from the end of the AHP pulse to the start of signal acquisition. The minimum T_E was 18 ms (corresponding to τ_cp = 3 ms) due to additional time needed for slice selective and spoiling gradients. For T₂ measurements, T_E was increased by adding free precession delays evenly distributed around the six AFP pulses (1).

In CP-LASER (Figure 1B), T_E was increased by adding AFP pulses rather than free precession delays. τ_cp was kept at 3 ms as in LASER. No gradient spoilers were used during the CP pulse train, and the CP pulses were phase cycled according to the MLEV scheme (20).

The T_2ρ-LASER sequence (Figure 1C) was similar to CP-LASER except that there was no delay between the CP pulses such that τ_cp was 1.5 ms (equal to the duration of one AFP pulse).

For LASER, ¹H spectra were acquired with eight T_Es: 18, 25, 32, 46, 60, 100, 200 and 400 ms. For CP-LASER and T_2ρ-LASER, CP train lengths (n_mlev) of 8, 16, 32, 64 and 128 AFP pulses were used, corresponding to T_Es of 42, 66, 114, 210 and 402 ms for CP-LASER and 30, 42, 66, 114 and 210 ms for T_2ρ-LASER. These T_Es correspond to the minimum T_E used in LASER (i.e., 18 ms) plus the length of the CP pulse train (i.e., n_mlev × τ_cp).

Water suppression was achieved using VAPOR (21). Spectra were acquired with T_R of 4 s, a spectral width of 10 kHz and 10,000 data points. Each FID was individually saved for shot-to-shot B₀ frequency correction. A non-suppressed water spectrum at each T_E was also acquired for eddy current correction. For each pulse sequence, macromolecule spectra were acquired using the inversion-recovery technique (22) for all T_Es except for T_Es > 200 ms where macromolecule signal was negligible.

All in vivo ¹H spectra were analyzed using LCModel version 6.1-4A (Stephen Provencher Inc., ON, Canada (23)). No baseline correction, zero-filling or apodization functions were applied to the in vivo data prior to the analysis. The spectra were fitted over the 0.5 to 4.2 ppm range. The model basis spectra for the various T_Es used in this study were simulated using home-written programs based on density matrix formalism in Matlab (The MathWorks, Inc., Natick, MA, USA) (24) using measured and published chemical shifts and J-coupling values (25). The spin evolution occurring during the adiabatic RF pulses was also taken into account in the simulations. Separate basis spectra were generated for the singlet and multiplet signals of NAA, and the CH₂ and CH₃ groups of creatine and phosphocreatine (6). The signal from the methylene protons of tCr present in the macromolecule spectra (due to their short T₁ relaxation time) was removed in post-processing using HSVD in Matlab.

The T₂ values were determined by fitting the metabolite signals obtained from LCModel analysis as a function of T_E using a two-parameter, mono-exponential decaying function in Matlab. Only metabolites with Cramér-Rao Lower bounds (CRLBs) < 50% at all T_Es were considered (supplementary material), and T₂ values are reported for metabolites for which fits with goodness of fit (R²) above 0.88 were obtained. The measured transverse relaxation times for each sequence are denoted by T_2f (free precession), T₂^† and T_2ρ for LASER, CP-LASER and T_2p-LASER, respectively. The two-tailed, unpaired student's t-test was used to compare T₂s between pulse sequences.

RESULTS

Spectral appearance

In vivo ¹H NMR spectra acquired at various T_Es with LASER, CP-LASER and T_2ρ-LASER sequences in the rat brain are shown in Figure 2. In LASER, singlets showed signal reduction with increasing T_E due to T₂ relaxation while J-coupled metabolites underwent both J-evolution and T₂ relaxation, which is apparent mainly in the 2.1 to 2.5 ppm (glutamate and glutamine) and 3.4 to 3.6 ppm (mIns) regions. In CP-LASER and T_2ρ-LASER spectra, in contrast, the intensities of these multiplets were preserved at relatively long T_Es: glutamate and glutamine resonances were still visible at 114 ms while mIns began to undergo noticeable J-modulation only around 210 ms.

Figure 2.

In vivo ¹H NMR spectra measured with LASER, CP-LASER and T_2ρ-LASER sequences from the same animal with different T_Es (T_R = 4 s, 128 averages were acquired for all spectra except for T_E = 400 ms in LASER where 192 averages were acquired). The T_2ρ-LASER spectrum at 402 ms was only acquired in this animal to show that even at such long T_E several resonances in the 3 to 4.2 ppm range could be still observed compared to LASER and CP-LASER spectra.

Spectra obtained with CP-LASER and T_2ρ-LASER were remarkably similar at the same T_E (Figure 2). However, small differences could be observed especially at long T_Es (Figure 3). For example, the signal intensity of mIns at T_E of 210 ms was lower with CP-LASER than with T_2ρ-LASER despite having very similar spectral pattern (Figure 3a). In addition, glutamate exhibited different J-modulation with CP-LASER than T_2ρ-LASER, and its intensity was higher with T_2ρ-LASER. At T_E of 402 ms, no mIns signal was observed with CP-LASER due to the J-modulation while under T_2ρ-LASER mIns did not undergo any J-evolution (Figure 3b). These differences can be attributed to the different number of CP pulses and the different τ_cp delay between the CP pulses (3 ms versus 1.5 ms) in both sequences. The J-modulation is slowed down further in T_2ρ-LASER compared to CP-LASER, especially for strongly coupled spin systems. Simulated spectra were in excellent agreement with the spectra measured in vivo (Figure 3).

Figure 3.

Comparison of LASER, CP-LASER and T_2ρ-LASER spectra (128 averages, a 1.5 Hz Gaussian line broadening was applied prior to FT) measured at (a) T_E = 200 ms and 210 ms and (b) T_E = 400 ms and 402 ms from the same animal. In (a), the simulated spectra are scaled by T₂ relaxation (amplitude matched to in vivo spectra). The observed differences in intensities between CP-LASER and T_2ρ-LASER sequences are consistent with the density-matrix simulated NMR spectra. In (b), the simulated mIns spectra are shown with the same scaling adjusted so the intensity of mIns in T_2ρ-LASER simulated spectrum matches the intensity of mIns in the T_2ρ-LASER in vivo spectrum. Therefore the simulated spectra represent the relative signals intensity of mIns that would be expected if T₂ was the same with LASER, CP-LASER and T_2ρ-LASER. With the same T₂ relaxation, the peak at 4.05 ppm would have almost identical amplitude with LASER and T_2ρ-LASER. However, it is apparent that in vivo this peak is much smaller in LASER spectrum compared to T_2ρ-LASER spectrum providing direct visual evidence that the T₂ of mIns is longer with T₂-LASER compared with LASER.

The simulated spectra are shown line-broadened to match in vivo spectra (8 Hz exponential line-broadening). Glu = glutamate, mIns = myo-inositol.

Macromolecule resonances were also affected by the CP pulse train. For instance, the signal intensities for the 0.91 ppm and 1.7 ppm resonances were higher with CP-LASER and T_2ρ-LASER compared to LASER at similar echo times (Figure 2). However, no macromolecule signal was observed for any sequences at an echo time of 200 ms and longer.

Effect of CP pulse train on J-modulation

Under CP conditions, signal loss due to J-modulation was reduced much more significantly in strongly coupled spin systems than in weakly coupled spin systems as shown previously with simulations for alanine and taurine (26). This is illustrated in Figure 4 which shows simulations of total signal integral (without taking into account T₂ relaxation) as a function of T_E for glutamate (a strongly coupled spin system) and lactate (a weakly coupled spin system) for all three pulse sequences.

Figure 4.

Effect of CP pulse trains on the J-evolution of glutamate (strongly coupled system) and lactate (weakly coupled system). Signal integral (without taking into account T₂ relaxation) which corresponds to the amplitude of the first point of the simulated FID normalized to 1 at T_E of 18 ms is shown as a function of T_E for LASER (blue line), CP-LASER (black line) and T_2ρ-LASER (red line) sequences at 9.4 T. Inserts show the spectral pattern of glutamate at T_E = 66 and 126 ms, and lactate at T_E = 150 and 300 ms for the three sequences. Although the signal integral of glutamate with LASER is almost zero at T_E = 66 ms, there is sufficient signal intensity for precise LCModel fitting. At T_E = 126 ms, the signal intensity (without T₂ relaxation) of glutamate with LASER is 2.5 times higher than that with T_2ρ-LASER. It is possible that for certain T_Es and certain metabolites, LASER sequence might have more signal than CP-LASER or T_2ρ-LASER.

The simulated spectra are shown line-broadened (10 Hz exponential line-broadening).

For glutamate, simulations show that J-modulation is substantially reduced with CP-LASER and T_2ρ-LASER compared to LASER, which is particularly apparent for T_E < 150 ms. For lactate, in contrast, J-modulation is not significantly slowed by the CP train for T_E < 100 ms, although it differs for each sequence at longer T_E. The signal integral of lactate with LASER follows a sinusoidal modulation as T_E increases, with the first minimum occurring at ~144 ms (i.e., 1/J) and the next maximum at ~288 Hz (i.e., 2/J) as expected for a weakly coupled ½ spin system. With the CP pulse train, the lactate integral is not refocused at multiples of 1/J.

This could also be observed in vivo (Figure 2): resonances of weakly coupled spins such as the methyl protons of lactate at 1.32 ppm and the NAA multiplet in the 2.49–2.67 ppm region showed noticeable J-modulation at much shorter T_E than resonances of strongly coupled spins such as glutamate, taurine and mIns, with CP-LASER and T_2ρ-LASER sequences.

Finally, simulations in Figure 4 also confirm that, while CP-LASER and T_2ρ-LASER behave very similarly at shorter T_E, significant differences in J-modulation become apparent at longer T_E, as can also be observed in experimental data (Figure 3).

Transverse relaxation times

The T₂ relaxation times of metabolites measured using all three pulse sequences are given in Table 1. T₂s of all measured metabolites were significantly longer with CP-LASER and T_2ρ-LASER compared to LASER. However, no significant difference was observed in measured T₂s between CP-LASER and T_2ρ-LASER sequences for any metabolite.

Table 1.

T₂ relaxation times (mean ± SD) measured with LASER, CP-LASER and T_2ρ-LASER sequences and T₂ ratios for CP-LASER and LASER and T_2ρ-LASER and LASER obtained in the rat brain (n = 6) in vivo at 9.4 T. T₂s of all measured metabolites were significantly longer (P < 0.05) with CP pulse sequences compared to LASER. T₂ of water was also significantly longer with CP-LASER (P < 0.01) and T_2ρ-LASER (P < 0.001) compared to LASER. Between CP-LASER and T_2ρ-LASER sequences, the T₂s of metabolites were not significantly different (P > 0.05) except in the case of water (P = 0.04). The high SDs observed for tCho and mIns were due to their T₂ being longer than the maximum T_E used in this study. The goodness of fit, R² was > 0.9 for all metabolites except for the multiplet of NAA which was 0.88 with LASER.

Spin systems	Compound	Group	LASER T2f (ms)	CP-LASER T2† (ms)	T2ρ-LASER T2ρ (ms)	T2†/T2f	T2ρ/T2f
Non-coupled and weakly coupled	Water	H2O	44 ± 2	48 ± 2	50 ± 2	1.09 ± 0.07	1.16 ± 0.07
	NAA	2CH3	321 ± 30	456 ± 26	449 ± 32	1.42 ± 0.16	1.40 ± 0.17
	NAA	3CH2	157 ± 41	187 ± 16	167 ± 25	1.19 ± 0.33	1.06 ± 0.32
	tCr	N(CH3)	170 ± 12	224 ± 7	231 ± 16	1.32 ± 0.10	1.36 ± 0.13
	tCr	2CH2	146 ± 16	175 ± 9	183 ± 20	1.20 ± 0.15	1.25 ± 0.20
	tCho	entire molecule	445 ± 67	523 ± 68	ND	1.17 ± 0.23	-
Strongly coupled	Glu	entire molecule	70 ± 14	296 ± 44	270 ± 21	4.20 ± 1.02	3.84 ± 0.80
	Gln	entire molecule	56 ± 12	116 ± 7	126 ± 19	2.08 ± 0.45	2.26 ± 0.58
	mIns	entire molecule	161 ± 26	584 ± 94	600 ± 90	3.62 ± 0.82	3.72 ± 0.82
	taurine	entire molecule	162 ± 28	355 ± 17	343 ± 27	2.19 ± 0.39	2.11 ± 0.40

Singlet resonances showed a significant increase in T₂s with CP-LASER compared to LASER, ranging from 17% for tCho to 42% for NAA (Table 1). Similarly, the multiplet of NAA in the 2.47 to 2.67 ppm region showed a 6-19% increase in T₂ with CP-LASER and T_2ρ-LASER compared to LASER (Table 1). In comparison, the T₂ of water increased by 9% with CP-LASER and 16% with T_2ρ-LASER compared to LASER.

For strongly coupled systems, the increase in measured T₂s with CP-LASER and T_2ρ-LASER compared to LASER was much larger. For example, T₂ of glutamate increased from 70 ms with LASER to 296 ms with CP-LASER. The increase in T₂ with CP-LASER compared to LASER was about 2-fold for glutamine and taurine and about 4-fold for mIns and glutamate (Table 1). mIns exhibited the longest T₂ under CP condition (600 ms). This lengthening of T₂ is readily apparent when comparing the in vivo and simulated spectra of mIns and its resonance at 4.05 ppm (Figure 3B). Similarly, the prolonged T₂ was confirmed directly for taurine using phantom data acquired with the three sequences (supplementary material).

The T₂ of tCho could not be determined under T_2ρ-LASER due to the strong cross-correlation between choline compounds and phosphorylethanolamine in the LCModel fit. In addition, the T₂ of lactate could not be determined reliably with CP-LASER and T_2ρ-LASER due to the lower signal integral of lactate with CP trains compared to LASER at most T_Es (Figure 4).

DISCUSSION

In this study, we report for the first time quantitative measurement of T₂ relaxation times for multiple J-coupled metabolites in the brain under CP conditions. While all metabolites had longer T₂ with CP-LASER and T_2ρ-LASER compared to LASER, the increase in T₂ was much more pronounced for strongly coupled spins (up to 4-fold longer) than for non-coupled (singlet) and weakly coupled spins. These findings provide new insights into the relaxation pathways in coupled spin systems and have significant implications for accurate quantification of MR spectra acquired at high field.

Comparison of T₂ measured with LASER, CP-LASER and T_2ρ-LASER

T₂s measured for non-coupled (singlet) spins including water were longer when measured with CP-LASER and T_2ρ–LASER sequences compared to LASER. However, the increase in T₂ was smaller for water than for other singlets such as the CH₃ group of NAA or the CH₃ and CH₂ groups of tCr. Similar observations have been previously reported for water, NAA and tCr in the human brain (5). For water which has a larger apparent diffusion coefficient (ADC) than metabolites in addition to rapidly exchanging protons, the increase in T₂ was smaller. This suggests that the diffusion component might not be efficiently refocused by the CP pulse trains (5). In contrast, NAA and tCr singlets have non-exchangeable protons and lower ADC than water, so that the CP pulse train suppresses the diffusion component efficiently, resulting in higher increase in T₂ under CP train.

Although the molecular structures of glutamate and glutamine are similar, these metabolites seem to experience different relaxation pathways under the CP pulse train. This is evidenced by the much higher increase in T₂s for glutamate (4-fold) compared to glutamine (2-fold) with CP-LASER and T_2ρ-LASER compared to LASER. This different gain may be explained by the higher diffusion coefficient for glutamine compared to glutamate as previously reported in the rat brain (27). With a high diffusion coefficient, CP pulse trains may not refocus the diffusion component of T₂ relaxation as efficiently (5).

Strongly coupled spins such as glutamate, taurine, and mIns exhibit a higher gain in T₂s under the CP pulse train compared to non-coupled spins although their diffusion coefficients are closer to that of singlet resonances. This suggests that the prolonged T₂ values cannot be solely attributed to suppression of the diffusion term since both non-coupled and J-coupled spins experience this relaxation mechanism. However, in strongly J-coupled systems, the additional increase in T₂ during a CP sequence can be attributed to refocusing of cross-correlation or interference effects (28) between different dipole-dipole interactions (29).

Finally, T₂s measured with CP-LASER and T_2ρ-LASER were very similar, although differences were observed in the J-modulation pattern at longer T_E. The similarity in T₂ values can be attributed to the relatively short τ_cp of 3 ms used in CP-LASER, resulting in minimal dephasing of magnetization during the interpulse interval (30). Longer τ_cp delays (> 3 ms) in CP-LASER would result in shorter T₂^† compared to T_2ρ. At very long τ_cp, T₂^† would approach that of T_2f as previously reported (16,30,31).

The transverse relaxation times are dependent on magnetic field strength and the properties of the pulse sequence utilized (i.e., RF pulse patterns and timings) such that T₂ values and T₂ gains observed in this study at 9.4 T might not hold true for other sequences and other timings. For example, the T₂s of singlets (e.g., NAA, tCr and tCho) measured with LASER in the present study were consistent with a previous rat study using a similar pulse sequence (7) but were longer (between 30% to 250% longer) than those reported with PRESS or SPECIAL sequences (32,33) at 9.4 T. Additionally, T₂ values measured with LASER but not with CP-LASER and T_2ρ-LASER can be affected by the chemical shift displacement errors. In the present study, the largest chemical shift displacement error of 7% would be observed for lactate due to the large bandwidth of the AFP pulses. However, for the strongly coupled metabolites such as glutamate, glutamine, taurine and mIns the chemical shift displacement error would be smaller (4%, 4%, 0.4%, and 2%, respectively).

Comparison of effect of CP train in human brain and rat brain

The increase in T₂ under CP conditions (i.e., T₂^†/T_2f) is dependent on B₀, as reported previously (5), because the dynamic dephasing (i.e., diffusion and exchange) component of transverse relaxation increases with B₀. The T₂s measured for non-coupled spins (NAA, tCr and tCho) with CP-PRESS (12) in human brain at 1.5 T were only slightly longer compared to values obtained with the PRESS sequence (34). At higher fields, however, the increase in T₂ with CP-LASER compared to PRESS was larger. The T₂ of NAA singlet in human brain was 1.7 times longer at 4 T and 2.2 longer at 7 T with CP-LASER compared to PRESS (5), and 1.5 times longer at 4 T with CP-LASER (5) compared to LASER (6).

In the present study in rat brain, the increase in the T₂ of NAA singlet at 9.4 T with CP-LASER compared to LASER was only a factor 1.4, which was lower than the increase observed in human brain at 4 T. Therefore the increase in T₂ with CP-LASER compared to LASER appears smaller in rat brain than in human brain at a given B₀. This is likely due to smaller microsusceptibility effects in rat brain, resulting in smaller B₀ dephasing effects (35,36). A similar dependence of T₂ on microsusceptibility and magnetic field can be expected for J-coupled spins.

Effect of RF pulse modulation on T₂ relaxation times

The increase in T₂ during CP pulse trains depends both on the amplitude and frequency-modulation functions of the AFP pulses used (37). It was experimentally demonstrated on phantom using ¹H MRS that water exhibits longer T_2ρ relaxation times under HS4 AFP pulses compared to HS1 AFP pulses with similar pulse duration and B₁ field (31,38,39). Similar findings were recently reported for the CH₃ group of NAA and tCr in the human brain in vivo at 4 T (39). This dependence of T₂ on modulation functions is the result of spins experiencing different dipole-dipole and/or chemical exchange interactions during the different type of pulses (37) and forms the basis of a new MRI contrast (31). In the present study, HS1 pulses were used for both CP train as well as in LASER. Even larger increases in T₂ can be expected if HS4 pulses are used.

Absolute quantification

Sequences such as LASER and semi-LASER are advantageous at high field because AFP pulses can be lengthened while retaining a large refocusing bandwidth, which minimizes chemical-shift displacement artifacts. Even though LASER and semi-LASER have longer T_E than STEAM, the effect of this longer T_E is mitigated by reduced J-modulation and increased T₂. In fact, it has been shown that LASER and semi-LASER spectra can be very similar in appearance to short echo-time STEAM spectra while providing 2-fold higher signal-to-noise.

At high magnetic fields, the minimum achievable T_E generally becomes longer due to increased RF power requirements, which leads to longer RF pulses. At the same time, T₂s relaxation times become shorter at high field (36). Therefore, while often neglected at lower field, signal loss due to T₂ relaxation may not be negligible at high field, especially in human brain.

Correction for T₂ relaxation requires measurement of the T₂s of metabolites. Our results show that the common practice of measuring T₂ by inserting free precession delays does not provide accurate T₂ values for CP-like sequences. For this type of sequences such as LASER and semi-LASER, T₂s must be measured while keeping τ_cp constant to take into account the effect of successive 180° pulses. The potential effect of successive 180° pulses on T₂ is important to keep in mind when performing absolute quantification.

CONCLUSION

This study reports for the first time quantitative measurements of T₂s for multiple J-coupled metabolites. It is shown that the use of a CP pulse train results in a much larger increase in T₂ for strongly coupled spins (up to 4-fold) than for non-coupled and weakly coupled spins. Therefore, pulse sequences with a succession of 180° pulses such as LASER are particularly advantageous for retaining signals from strongly coupled spins at relatively short T_E. If correcting for T₂ relaxation is needed in LASER and semi-LASER, care must be taken to use T₂ values measured under CP-like conditions.