Evaluation of optimal echo time for 1H-spectroscopic imaging of brain tumors at 3 Tesla

Abstract

Purpose: To compare the spectral quality of short echo time (TE) MR spectroscopic imaging (MRSI, TE = 30 msec) with long-TE MRSI (TE = 144 msec) at 3 Tesla in normal brain and tumor tissue.

Materials and methods: Spectroscopic imaging (chemical-shift imaging (CSI)) data of 32 patients with histopathological confirmed brain lesions were acquired at 3 Tesla (3T) using TEs of 30 msec and 144 msec. Tumor-relevant metabolites (trimethylamine (TMA), creatine compounds (tCr), and N-acetylated compounds (tNAA)) were analyzed with LCModel software, which applies prior knowledge by performing a frequency domain fit using a linear combination of model spectra.

Results: Short-TE spectra provided up to twice the signal-to-noise ratio (SNR) compared to TE = 144 msec. The estimated fitting error was improved up to 30% for TMA and tCr, but was slightly reduced (10%) for tNAA. Quantification in terms of absolute concentrations was consistent at both TEs.

Conclusion: Since other metabolites observable at TE < 30 msec may be of diagnostic relevance, short-TE MRSI should be the preferred method at 3T for the evaluation of focal lesions in brain tissue; however, TE = 144 msec can serve as an option for MRS in regions with potential baseline problems.