The SNR is based on several variables, the most important of which are of course signal intensity (where Cr is often used as the reference) and the underlying noise of a spectroscopy experiment. The dependencies of signal intensity and noise can be divided into two categories:
1. Dependencies which cannot be affected by the user and which are given by fixed natural constants. The corresponding SNR shall be called SNRint.
2. Dependencies that can be altered by modifying the acquisition parameters. The corresponding SNR shall be called SNRExp.
The most important factors for SNRint in a 'H-MRS experiment are to be aware of are the number of protons contributing to the total signal N, the field strength of the static magnetic field B0 and the relaxation properties of a specific metabolite. With a direct, linear proportionality of SNRint to N and B0 these parameters define the intrinsic SNR available for the spectroscopy experiment, which means that an increase in B0 from 1.5Tto 3.0 T will theoreticallyboost the SNRby a factor of two. This achievable SNR will always be degraded by the natural phenomena of relaxation, expressed as an exponential signal decay after full excitation with the relaxation constants T2 and exponential return of the spin system into thermal equilibrium with the time constant T1.
For a given sample in a specific MR scanner, the value of SNRint is fixed, and its limitations cannot be overcome. Beside these factors, there are others which can be optimized by the user, and which contribute to the final SNR. The main parameters to be considered by the user are the type of sequence, the number of signal averages N, the sample volume (VOI), and the echo and repetition times (TE, TR).
The SNR gain obtained from the intensity increase from 1.5 to 3.0 T can for instance be used to decrease the acquisition volume by a factor of two. On the other hand, reduction of the acquisition volume by a factor of two at a constant field strength would require to in-
Fig. 6.3. Spectra acquired on the same phantom containing the main brain metabolites at 1.5 T (a) and at 3.0T (b) usingthe same sequence parameters (PRESS: TR: 2000 ms, TE: 35 ms), showing the increased SNR, the improved spectral resolution, in particular between 2 ppm and 2.6 ppm, and the different peak ratio between the main metabolites due to the different relaxation times of individual metabolites at different field strengths b a crease the number of averages by factor of four to maintain the SNR. In addition, T1 relaxation times increase with higher field strength, leading to increased signal saturation for a given repetition time, and T2 relaxation times decrease. Therefore, the theoretical doubling of SNR cannot be achieved, due to the use of repetition times (TR) in the order of the T1 decay times (and not infinitely long) and echo times (TE) in the order of the metabolite T2 decay times [19-22].
Comparison of different experimental settings thus requires careful analysis of all parameters to avoid errors and misinterpretations (Fig. 6.3).
Was this article helpful?