MRS Localization

If frequency-encoding cannot be used to determine spatial position, how do you localize an MRS signal?  

In MRS spatial frequency-encoding cannot be used since frequency differences must be reserved for identifying chemical shifts between molecular species. Several localization strategies are available: 1) Single voxel methods; 2) Multi-voxel methods; and 3) Surface coil methods.

Single Voxel Spectroscopy (SVS)

Single-voxel techniques use sequentially applied RF-pulses coupled with gradients in three orthogonal planes. The intersection of these planes defines a cube-shaped voxel as the source of the MR signal. 

Intersection of 3 orthogonal planes defines a cubic voxel for SVS.

Single Voxel Spectroscopy (SVS)

Three principal SVS methods are in common use, distinguished by the nature and timing of RF-pulses and the types of signal generated: ​

1. Point RESolved Spectroscopy (PRESS) is the most popular method for ¹H spectroscopy. It uses 3 RF-pulses (90º−180º−180º) and generates a spin-echo signal. 
2. Stimulated Echo Acquisition Mode (STEAM) uses three 90º-pulses and generates a stimulated echo. Because of its inferior signal-to-noise compared to PRESS, STEAM has progressively fallen out of favor for ¹H spectroscopy during the last decade for fields ≤ 3.0 T.
3. Image-Selected In vivo Spectroscopy (ISIS). Unrelated to the terrorist organization, ISIS is used primarily for ³¹P spectroscopy. Free induction decay signals are generated from 8 separate RF-pulse cycles, then added and subtracted in a certain order to define the volume of interest. 

(Multi-Voxel) Chemical Shift Imaging (CSI)

Chemical Shift Imaging (CSI) is the traditional name for a group of MRS techniques also referred to by the more modern term MR Spectroscopic Imaging (MRSI). Spectra can be simultaneously obtained in 1D (a column of voxels), 2D (a plane of voxels), or 3D (block of voxels). 

2-D Chemical Shift Imaging (CSI)

CSI phase-encoding can be combined with any type of excitation and signal generation method. In the simplest possible case, the entire volume could be excited with a non-selective RF-pulse, with sampling of an FID signal after each phase-encoding step. More commonly, excitation is performed using spatially selective RF-pulses and gradients similar to PRESS, STEAM, or ISIS. The use of multiple gradient steps means that CSI-methods (especially those in 3D) are time-intensive compared to SVS techniques. To speed data acquisition, various methods (including parallel imaging, reduced k-space sampling, and multi-band excitation) may be used. A relatively straightforward method to reduce imaging time is to acquire two or more echoes during each TR interval. This technique is known as Turbo Spectroscopic Imaging (TSI), with its obvious analogy to turbo spin echo (TSE) imaging.

Surface Coil Localization

Surface coil localization combining sensitive volume with 1D CSI

For multinuclear (i.e, non-¹H) spectroscopy, small surface coils tuned to the nucleus of interest are typically placed directly over the desired anatomy (heart, liver, muscle). Such coils have a conical receptive fields extending into tissue a distance approximately equal to their diameters. If nonselective RF-pulses are applied without spatial gradients, the resultant signal derives from the entire sensitive volume of the coil (i.e., a single giant hemispheric shaped voxel). If slice-selective RF-excitation or phase-encoding gradients are used, then more specific localization to disks, single, or multiple voxels can be obtained. ​

A summary of the major MRS localization methods, together with their advantages and disadvantages is provided in the Table below. More details about PRESS, STEAM, ISIS, and CSI can be found in dedicated Q&A's related to each technique.

Advanced Discussion (show/hide)»

Further comments about SVS and CSI methods

Each of the overlapping planes defining a voxel in SVS are generated by the simultaneous application of a slice-selective RF-pulse together with a magnetic field gradient. The thickness of each defining plane (d) is determined by the bandwidth (BW) of the pulse and the gyromagnetic ratio (γ) according to the equation: d = BW/γG. For example, if BW=2000 Hz, γ=43 MHz/T and G=5 mT/m, the plane thickness would be approximately 9 mm.

If the gradient strength (G) is the same in each direction and the gradients are orthogonal (mutually perpendicular), the resulting voxel would be a perfect cube. If gradient strength were changed for each direction, however, the sides of the voxel would differ and would become a rectangular prism ("cuboid"). Moreover, there is no absolute requirement that the planes be orthogonal; in this uncommon implementation the overlap volume would be a parallelepipid.

In general it is desirable to keep the RF-bandwidth (BW) as large as possible to minimize the so-called chemical shift displacement artifact, described more completely in a later Q&A. Water-fat chemical shift artifacts are well known in conventional MR imaging, resulting in anatomic misalignments and/or phase cancellations based on the slightly different resonance frequencies of water and fat protons (~3.5 ppm). In MRS, chemical shift differences also affect the presumed locations of other metabolites, causing them to be displaced relative to one another. In addition to spatial mismatch, chemical shift displacement also results in spectral contamination, where molecules outside a region may "bleed" into the defined voxel.

Although PRESS, STEAM, and ISIS are the most popular localization strategies for SVS, a fourth method is used at high and very high fields (3T - 7⁺T) where local coils are more commonly employed for RF transmission. LASER (Localization by Adiabatic SElective Refocusing) and its streamlined variant "semi-LASER", resemble PRESS in that they use a 90°-excitation pulse followed by 2-3 pairs of adiabatic 180°-pulses. The adiabatic pulse pairs, with their high bandwidths and more uniform excitation profiles, reduce chemical shift artifacts and are more tolerant to B₁ inhomogeneities than PRESS or STEAM methods.

Further details about PRESS, STEAM, ISIS, LASER and CSI methods can be found in the references and in specific Q&A's devoted to each technique separately.

Further comments about Surface Coil Methods

Surface-coil based strategies were among the earliest approaches for MRS spatial localization. These methods became less important as superconductive magnet design evolved with improved homogeneity, eddy current compensation, and stable/accurate gradients that could be rapidly switched.

The most straightforward use of surface coil localization is to use the entire sensitive volume of the surface coil as a giant single voxel. This whole-volume method is still used clinically for ³¹P MRS of skeletal muscle, where a small (~7.5 cm) coil is placed over the muscular mass of the leg or arm with spectra recorded before, during, and after exercise. The usual technique is a pulse and acquire free induction decay (FID) sequence from the entire coil. A similar strategy can be applied to the liver and heart.

In the 1980s magnetic field gradients applied perpendicular to the surface coil were employed to give somewhat better spatial resolution. One of the most popular methods was DRESS (Depth RESolved Spectroscopy), which allowed parallel disk-like slices to be obtained at various distances below the coil. DRESS, although historically important, has now been largely supplanted by chemical shift imaging (CSI) methods using phase-encoding for slice localization, but may occasionally still be encountered in hybrid forms for specific research applications.

An interesting class of surface coil-based localization strategies revolve around the use of radiofrequency field (B₁) gradients. Unlike the much more familiar switched B_o magnetic field gradients used for traditional MRI and MRS, B₁-gradient techniques exploit localized differences in excitation efficiency, flip angle, and phase shifts from one or more surface coils to provide spatial encoding. Two of the most widely known methods are the composite/depth pulse method EXORCYCLE and Rotating Frame Zeugmatography (RFZ). Although not in common use for MRS at present, I recommend keeping an eye out for these B₁-gradient strategies in the future.

---

References 
     Aue WP, Müller S, Cross TA, Seelig J. Volume-selective excitation. A novel approach to topical NMR. J Magn Reson 1984; 56:350-354. (First published example of SVS using 3 pulses with orthogonal gradients).  
     Bertoldo D, Watcharakorn A, Castillo M. Brain proton magnetic resonance spectroscopy. Introduction and overview. Neuroimag Clin N Am 2013; 23:359-380.
     Bottomley PA. Spatial localization in NMR spectroscopy in vivo. Ann NY Acad Sci 1987; 508:333-348. (Excellent historical review of major MRS localization methods)
     Bottomley PA. Selective volume method for performing localized NMR spectroscopy. US Patent #4,480,228 (approved 30 Oct 1984). (First description of the method later to known as PRESS).
      Frahm J, Bruhn H, Gyngell ML, et al. Localized high resolution proton NMR spectroscopy using stimulated echoes: initial applications to human brain in-vivo. Magn Reson Med 1988; 9:79-93.
     Hajek M, Dezortova M. Introduction to clinical in vivo MR spectroscopy. Eur J Radiol 2008; 67:185-193.
     Kimmich R, Hoepfel D. Volume selective multipulse spin-echo spectroscopy. J Magn Reson 1987; 72: 379-387.
     Keevil SF. Spatial localization in nuclear magnetic resonance spectroscopy. Phys Med Biol 2006; 51:R579-636. (good, technical review of a wide range of methods, many of historical interest only)
     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.
     Ordidge RJ, Connelly A, Lohman JAB. Image selected in vivo spectroscopy (ISIS). A new technique for spatially selective NMR spectroscopy. J Magn Reson 1986; 66:283-294. 
     Öz G, Alger JR, Barker PB, et al. Clinical proton MR spectroscopy in central nervous system disorders. Radiology 2014; 270:658-679.
     Posse S, Otazo R, Dager SR, Alger J. MR spectroscopic imaging: principles and recent advances. J Magn Reson Imaging 2013; 37:1301-1325.
     Shankar RV, Chang JC, Hu HH, Kkodibagkar VD. Fast data acquisition techniques in magnetic resonance spectroscopic imaging. NMR Biomed 2019; 32:e4046. (Good insight into current and future techniques to speed up MRSI)
     Skotch A, Jiru F, Bunke J. Spectroscopic imaging: basic principles. Eur J Radiol 2008; 67:230-239.

---

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 ↑