STEAM

How does the STEAM method for MR Spectroscopy work and when should it be used?  

STimulated Echo Acquisition Mode (STEAM) is a spectroscopic technique using three slice-selective 90º-pulses applied concurrently with three orthogonal gradients (x, y and z). The STEAM signal is a stimulated echo (STE) derived only from protons that have experienced all 3 RF-pulses. These protons are located in a cuboid-shaped voxel where the three planes overlap. 

Simplified diagram of the STEAM SVS sequence for MRS.
Methods for water and fat suppression are described in separate Q&A's.

The time of appearance of the STE depends on the spacing of the three RF-pulses. If the first two pulses are separated by time TE/2, the peak of the STE will occur precisely at TE/2 after the third RF-pulse. The interval between the second and third pulses, TM, is called the mixing time, which is usually kept at a minimum. During this period the magnetization is "stored" along the z-axis and does not undergo T2 decay. Thus the echo time (TE) for the sequence is defined as TE/2 + TE/2 and does not include TM. 

STEAM holds several advantages compared to PRESS. First, the sequence TE can be made very short (down to ~7 msec in practice), allowing detection of short T2 metabolites. Secondly, the exclusive use of 90º- (rather than 180º-) pulses allows for sharper slice profiles (→ better voxel edge definition), higher bandwidth (→ less chemical shift displacement artifact), and lower tissue energy deposition (→ smaller SARs).  

Notwithstanding these advantages, STEAM has a major signal-to-noise penalty based on the fact that stimulated echoes (instead of a spin echoes) are used. At the same TR's and TE's, the maximum signal from STEAM is only half as large as from PRESS. For this reason alone, STEAM has continually lost popularity over the last decade, especially for ¹H spectroscopy at 3.0T and below. We no longer use it in our clinical practice except for estimation of hepatic fat fraction in ¹H-liver MRS.

Advanced Discussion (show/hide)»

A more detailed look at the STEAM pulse sequence

Stimulated echoes are a particular type of MR signal that results from a series of three or more RF-pulses. Details of STE formation have been provided in an earlier Q&A and the reader is advised to review this topic before proceeding.

In brief, STE's are generated by the re-tipping of residual tissue magnetization into the transverse plane that has been "stored" in the longitudinal direction by prior pulses. In the STEAM sequence illustrated above, the first 90º-pulse tips the tissue magnetization into the transverse plane, while the second 90º-pulse returns some of it back into the longitudinal direction for "storage" until the final RF pulse arrives. This third 90º-pulse, acting in concert with the first two, tips magnetization back into the transverse plane with production of the STE. 

The full STEAM pulse sequence contains some additional features omitted for clarity in the simplified diagram above. Below is a more complete timing diagram as might be implemented on a clinical scanner. Many vendor-specific variations exist, especially with regard to crusher gradients, but the diagram below can be considered representative.

More detailed STEAM timing diagram

Each 90°-pulse is applied with an imaging field gradient and simulates a plane of protons. In the diagram above, these gradients are denote Gz, Gy, and Gx, colored pink, peach, and red above. The thickness and location of each plane is determined by the strength of each gradient together with the center frequency and bandwidth of the RF-pulse. 

Only protons in lying at the intersection of the 3 crossing planes experience all 3 RF-pulses and generate the STEAM echo. The selected planes are usually orthogonal (i.e., mutually perpendicular), so the resultant voxel has a cuboid shape. The sides of this voxel need not be parallel to the scanner's x-, y-, and z-axes, but can be rotated into any plane by applying two or more gradients simultaneously with each RF-pulse. The rotated gradients are nearly always orthogonal to one another but do not have to be; if not, the resultant voxel becomes a parallelepipid rather than a cuboid.  

The STEAM echo is sampled without the use of a readout gradient (as would be done in a conventional imaging experiment). This is because frequency differences in the signal must be used to calculate chemical shifts rather than being used for spatial encoding. The STEAM echo is somewhat asymmetric, and often only the second portion is sampled as an FID before being Fourier transformed to produce a spectrum. 

In addition to the final STEAM echo, two intermediate MR signals are generated (an FID and SE) that are not separately sampled. The FID results from the action of the initial 90°-pulse that tips magnetization into the transverse plane. The SE a (Hahn) spin-echo that results from the combined effect of the first two pulses. Note that this echo arises from a column of voxels along the intersection of the first two planes, since only protons in these voxels have experienced both RF-pulses.   

Crusher gradients (blue) are required to eliminate spurious signals from contaminating the STEAM spectrum. The size, location, and timing of these gradients varies among vendors, but a typical configuration is shown above. Crusher gradients are intended to destroy unintended FIDs and echoes arising outside the volume of interest. In PRESS these spurious signals result primarily from imperfect RF-pulse slice profiles.  

In STEAM, however, even more coherence pathways exist that must also be disrupted. As described in a prior Q&A, in addition to the FID, SE, and STE shown, a train of 3 RF-pulses generates three additional small echoes (not pictured) after the third RF-pulse. These extra echoes may interfere with the detection of the desired STE and need to be proactively suppressed. 

Interesting phenomena occur during the mixing time (TM) interval. Recall that the first 90º-pulse has tipped initial tissue magnetization into a transverse plane. The mixing time interval begins immediately after second 90º-pulse returns the residual transverse magnetization (i.e., that which has not been lost by T2 decay) back along the longitudinal (z-) direction.

Spins returned to the longitudinal axis have no transverse components, and hence do not undergo T2 decay, no matter how long the TM-interval may be. This is why the effective echo time (TE) of the STEAM sequence does not depend on TM. T1-relaxation does occur during this the TM interval, however. Although TM is usually kept at a minimum value, it can be lengthened and used to manipulate signals of metabolites on the basis of their T1 relaxation times and coupling constants. A slightly longer TM can also be used reduce eddy current artifacts. Finally, if TM is sufficiently long, a water suppression pulse may be added in this location to further improve spectral quality. 

---

References
     Frahm J, Merboldt K-D, Hänicke W. Localized proton spectroscopy using stimulated echoes. J Magn Reson 1987; 72:502-508.
     Klose U. Measurement sequences for single voxel proton MR spectroscopy. Eur J Radiol 2008; 67:194-201.
     Moonen CT, von Kienlin M, van Zijl PC, et al. Comparison of single-shot localization methods (STEAM and PRESS) for in vivo proton NMR spectroscopy. NMR Biomed 1989; 2:201–207. 
     Thompson RB, Allen PS. Response of metabolites with coupled spins to the STEAM sequence. Magn Reson Med 2001; 45:955–965.    

---

Related Questions
     What is a stimulated echo? 
     Can you explain how PRESS works and why is it the most popular MRS method?

---

←  Previous Question

Next Question  →

↑ Complete List of Questions ↑