MRS Parameters

How do you pick various imaging parameters (TR, TE, etc) for a brain MRS study?  

Before an MRS study can be performed, the operator must make decisions concerning various imaging parameters, including voxel size, repetition time (TR), echo time (TE), and number of excitations (NEX). These choices involve trade-offs in spatial and spectral resolution, imaging time, and visibility of different metabolites. These concepts will be illustrated using single voxel spectra from the parietal lobe of a normal subject performed with the PRESS technique. Although the results are specific to this technique and brain region, most of the concepts can be generalized to multivoxel imaging and other pulse sequences (STEAM, SE).

Voxel Size
​
Voxel size is the primary determinant of signal-to-noise (SNR) in both MRI and MRS. The effect is direct and linear: a voxel twice as large as another will have twice the SNR. In MRS the size of each peak is roughly proportional to the number of protons giving rise to that peak, so if the vertical axis remains a fixed dimension we would expect all peaks to become taller as voxel volume increases. 

---

In Single Voxel Spectroscopy (SVS) the three voxel dimensions are directly specified in mm. In Chemical Shift Imaging (CSI) voxel dimensions are more indirectly prescribed. Here the voxel dimension (d) in each phase encoding direction equals the field-of-view (FOV) ÷ the number of phase-encoding steps (Np).

Number of Excitations (NEX)

The number of excitations (NEX), known on some systems as the number of signals averaged (NSA), is an important determinant of SNR. For statistical regions discussed in a prior Q&A, SNR is generally proportional to the square root of the number of signals averaged (√NEX). So if NEX is quadrupled (x4), the SNR is doubled (x2). This can be appreciated in the PRESS spectra below obtained at TR = 1500 and TE = 144 ms. Note that the peaks are sharper and the baseline noise much reduced in the NEX = 256 image compared to the NEX = 8 image.

Imaging time is directly proportional to NEX, so even though better quality spectra are obtained by increasing NEX, a time penalty results. To keep imaging times within "reasonable" ranges (i.e., 5-15 minutes), typical NEX values might be 1-2 (for 3D-CSI), 2-8 (for 2D-CSI), and 64-128 (for SVS) as an acceptable tradeoff between time and quality.

Repetition and Echo Times (TR & TE)

Just like whole organs, individual brain metabolites have measurable T1 and T2 relaxation times. These differ somewhat by cerebral region, subject age, and field strenth. Average values measured at 3T for several common metabolites are shown in the table (right). 

As with conventional MR imaging, T2 values define the decay rate of the transverse magnetization. At short TE's (as in the TE = 30 msec spectrum below left), all metabolic peaks will be at their maximum values because they have had little time to undergo T2 decay. Additionally, the damping effects of J-coupling modulations are minimized. Metabolites with short T2 values (like glutamate and choline) will be relatively advantaged. Especially prominent at short TE values are the broad macromolecular peaks producing the prominent undulating baseline. For these multiple reasons a short TE sequence is essential for any brain ¹H MRS study.

By comparison, the medium and long TE examples (TE = 144 and 288 msec) have much more level baselines free of macromolecular undulations. The three major peaks (NAA, Cho, and Cr) are well seen, but the smaller peaks progressively disappear. Lactate, if present, would appear as an inverted doublet on the TE = 144 spectrum but be upright at TE = 288. For most brain studies only the TE = 144 spectrum is obtained. Because of progressive T2 decay, the absolute height of all peaks are reduced with increasing TE, and the noise level is increased.

Repetition time (TR) controls T1-weighting. When TR is much shorter than the T1 of a metabolite, longitudinal magnetization does not fully recover and the peak height is reduced. (Note how NAA, with its long T1, is relatively short on the TR = 1500 spectrum).  

Ideally, better spectra would be produced for all metabolites using very long TR values. However, imaging time directly related to TR, so in most cases this penalty is not deemed worth the benefit. In general clinical practice for both SVS and CSI spectroscopy, TR values in the 1500 − 2000 msec range are typically chosen as a compromise.

Advanced Discussion (show/hide)»

Then number of excitations (NEX) for spectral signal averaging is generally a power of 2 (e.g., 2, 8, 64, 128). As with conventional imaging, higher values of NEX improve signal-to-noise and reduce motion artifacts. With multiple excitations, further improvement can be obtained through the use of phase cycling.

As a simple example of phase cycling, consider a NEX=2 study where the phase of the RF-pulse for the second acquisition is inverted relative to that used in the first acquisition. This process produces an inversion of the recorded signal between the two acquisitions. The second (inverted signal) is then subtracted from the first, resulting in a net addition of the two signals while cancelling out RF errors independent of the fixed phase shift.

The actual range of measured metabolite T1s and T2s is quite large among different studies, sometimes varying as much as ± 30% from the average values cited. There can be appreciable differences among brain regions as well. A strong field dependence is also noted, with T2 values at 7T much shorter than those at 3T in the table.

---

References
     Bertoldo D, Watcharakorn A, Castillo M. Brain proton magnetic resonance spectroscopy. Introduction and overview. Neuroimag Clin N Am 2013; 23:359-380.      
     Govindaraju V, Young K, Maudsley AA. Proton NMR chemical shifts and coupling constants for brain metabolites. NMR Biomed 2000; 13:129-153. (Extensive tables, spectra, and descriptions for just about every metabolite ever detected on a brain MRS study). 
     Li Y, Ozturk-Isik E, Lupo JM, et al. T1 and T2 metabolite relaxation times in normal brain at 3T and 7T. J Mol Imaging Dynam 2012, S1-5.
     Posse S, Otazo R, Dager SR, Alger J. MR spectroscopic imaging: principles and recent advances. J Magn Reson Imaging 2013; 37:1301-1325.
     Träber F, Block W, Lamerichs R, et al. ¹H metabolite relaxation times at 3.0 Tesla: Measurements of T1 and T2 values in normal brain and determination of regional differences in transverse relaxation. J Magn Reson Imaging 2004; 19:537-545.
     Zhu H, Barker PB. MR spectroscopy and spectroscopic imaging of the brain. Methods Mol Biol 2011; 711:203-226.

---

Related Questions
     Are stimulated echoes used for imaging?  I've never heard of that.  
     What is a stimulated echo?
     How does PRESS differ from STEAM? Which should I use for MRS?  ​

---

←  Previous Question

Next Question  →

↑ Complete List of Questions ↑