Implementation and Optimization of Magnetic Resonance Spectroscopic Imaging Pulse Sequences at High Field (9.4T)

Abstract

Introduction: Magnetic resonance spectroscopic imaging (MRSI) is a method that combines magnetic resonance spectroscopy (MRS) with magnetic resonance imaging (MRI), thereby allowing signal acquisition from multiple voxels in an imaging plane simultaneously. Due to its ability to rapidly and simultaneously assess metabolic characteristics of the sample non-invasively, MRSI has great potential for clinical applications and basic researches. However, there are several technical challenges that need to be addressed before the successful implementation of the method, which become far more challenging at high field due to severe B0 and B1 inhomogeneity. This work aims to determine the sequence of choice for MRSI at high field by implementing the localization by adiabatic selective refocusing (LASER) and the semi-LASER 1H-MRSI pulse sequences and comparing their performance with conventional methods such as slab selection with outer-volume suppression (OVS), and a PRESS-based and a STEAM-based localization MRSI pulse sequences at 9.4T. Methods: To evaluate the performance of the proposed pulse sequences, a metabolite phantom was made, which consisted of glutamine (Gln), N-acetylaspartate (NAA), lactate (Lac), phosphocreatine (PCr), taurine (Tau), and phosphocholine (PCh) at respective concentrations of 50, 50, 45, 11, 50, and 20mM. Additionally, a phantom, which has two compartments consisting of water and soy oil, was made to assess the extent of the signal bleeding with the sequences. All experiments were performed on a 9.4T animal MR scanner (Agilent Technologies, Santa Clara, USA). A volume coil was used for both RF transmission and signal reception (Agilent Technologies, Santa Clara, USA). The maximum gradient strength was 40 G/cm. The first- and second-order shimming were performed by using FASTMAP. The MRSI sequence parameters (repetition time =2000ms, echo time (TE)=11~21ms, 16 averages, field-of-view=8 X 8 X 2.0 mm3, 8 X 8 X 1 matrix size, slices=1, spectral bandwidth=5000Hz, data points=2048, pulse duration = 1000ms (excitation) and 1400ms (refocusing), and bandwidth of refocusing pulse=5000Hz, receiver gain=2dB w/o water suppression, 30dB w/ water suppression) were the same for all experiments. Results: For conventional methods the poor localization performance was problematic even in phantom. For LASER, excellent localization profiles were achieved without the need of an OVS module. However, the long TE of 20.24ms with LASER can result in quantification errors , given that T2 relaxation and J-evolution of spin systems during this period can be substantial in vivo. For semi-LASER, the localization performance without OVS was not as effective as LASER. But in combination with OVS, the localization profile was comparable to that with LASER. Thus, in consideration of the shorter TE attainable, higher SNR, and reduced SAR with the sequence, semi-LASER may find applications instead of LASER. Conclusion: Given the excellent volume localization performance along with the shorter minimum TE attainable, higher SNR, and less SAR, semi-LASER may be the sequence of choice for in vivo MRSI at high field.