Introduction

In vivo metabolite relaxation reflects changes in cell-type specific cellular environments due to diseases.¹ The transverse in vivo metabolite relaxation time (T₂) are conventionally quantified by fitting the decay of signals acquired at different echo times (TEs). However, TE-dependent effects, such as diffusion, macromolecule, and J-coupling modulation, often produce sub-optimal conditions for T₂ measurements. For example, many spectral editing techniques require a fixed TE, making determining the T₂ of edited signal difficult. Recently, Li et al. proposes a novel multiple flip angle pulse-driven ratio of longitudinal steady states (MARzss) method for quantifying metabolite T₂ without varying TE.² This method uses long repetition time (TR), seven different flips angles (FA), and a linear equation to quantify T₂. Here we evaluate the feasibility of shortening the scan time of the MARzss method by >80% by reducing both TR and number of FAs.

Methods

The MARzss method measures the pulse driven longitudinal steady state magnetization (Mzss) at several different FAs and computes T₁/T₂ of metabolites using the following equation: $$ Rzss=\sqrt{\frac{T_{1}}{T_{2}}}\cdot tan(\frac{\alpha}{2}) [1] $$ Where $$$Rzss=\frac{M_{0}-M_{ZSS}}{\sqrt{M_0^2-(M_{0}-M_{ZSS})^{2}}}$$$, M₀ is the thermal equilibrium signal intensity, and α is the FA of a single RF pulse in the RF pulse train with interspersing field gradient (iPFG). Monte Carlo simulations were performed to evaluate the precision of linear fitting between seven-FA² and two-FA (0^o and 50^o) methods with the same total acquisition time. The average numbers for the seven-FA FIDs and for the two-FA FIDs were selected to generate the same total acquisition time for the two methods. Phantom studies were performed using a doped water-based sphere phantom on a 7T scanner (n = 5). The minimum TR for two-FA MARzss were computed using Bloch simulations. TR of 4.5s was used for FA =0, and TR of 3s with iPFG containing 170 RF pulses was used for FA = 50^o. T₁ was pre-determined using a conventional inversion recovery sequence. T₂ was solved using Eq. [1]. Reference T₂ values were provided by the originally proposed seven-FA acquisition with substantially longer TR². In vivo MRS data were acquired using a 32-channel head coil at 7T. Written informed consent was obtained from 2 healthy volunteers (2 males, 40 and 58 years old). VOI was placed in the occipital lobe. TR = 6.5s for FA = 0; TR = 2.5s for FA = 50^o. The total acquisition time was 14 mins for the seven-FA experiments² and 2.5 min for the two-FA experiments. Number of averages = 16 for each FA.

Results

Figure 1 shows the trajectory of longitudinal magnetization in the absence and presence of an iPFG. T₁ and T₂ values of human brain NAA³ were used in the Bloch simulations. For TR = 6.5s, FA = 0^o and TR = 2.5s, FA = 50^o, the deviation of the final M_z from steady state is < 0.1%, indicating negligible error due to T₁ saturation. Table 1 summarizes the mean and standard deviation of metabolite T₂s from fitting 100 sets of Monte Carlo simulations. For the same total scan time, the standard deviations of T₂ by the two-FA method are all substantially smaller than the corresponding seven-FA values. Phantom results are given in Figure 2. With only 18% of the scan time of the seven-FA measurement, the mean T₂s from two-FA measurements are similar to seven-FA measurements, and the standard deviations of T₂s from two-FA measurements are reasonably small (Table 2). Preliminary in vivo results are given in Figure 3 and the in vivo T₂ values were summarized in Table 2. Although there is a large variation in T₂ between the two subjects, the differences in T₂ values measured between the two-FA and seven-FA methods for the same subject are less than 5%.

Discussion and Conclusion

The Monte Carlo simulations, phantom and preliminary in vivo results indicate that the scan time of the MARzss method can be greatly reduced by using minimum TR and two-FA measurements. Because in linear regression the variance of the slope is inversely proportional to the variance of independent variables⁴, a two-FA method with the largest separation in FA produces the highest precision. However, when FA is too large, the signal intensity of weak signals (e.g., glutamate) may become too small to be reliably determined. Therefore, there is an optimal set of FA values for maximizing both the accuracy and precision of the T₂ measurement, which will be determined in our on-going study. Nevertheless, our current results have demonstrated that it is feasible to greatly shorten the overall scan time of the MARzss method without significantly compromising the quality of T₂ measurements.

Acknowledgements

This work was supported by the intramural research program of National Institute of Mental Health.