Sunday, 11 March 2012

Magnetic Resonance Spectroscopy (MRS)

Learning objectives

After reading this chapter, you should be able:
  • To set out the physical mechanisms used to differentiate metabolites according to resonance frequency
  • State the material conditions and optimization required to measure a spectrum
  • List the quality criteria of a spectrum
  • Present the different metabolites explored in brain MRS, their position on the spectrum and their interest
  • Explain the different stages to obtain a monovoxel spectrum
  • Describe PRESS and STEAM sequences and the influence of TE on the spectrum
  • Specify the adaptations required for spectroscopic imaging: spatial encoding, reduced acquisition time

Key points

  • The resonance spectrum identifies metabolites by:
    • Locating the peak(s), determined by chemical shift (ppm) resulting from the shield formed by the electronic cloud of hydrogen nuclei in the molecules
    • The same compound being characterized by several peaks (doublet, triplet) due to spin-spin coupling (or J coupling) phenomena: Lac (1.7 and 1.33 ppm)
  • Magnetic resonance spectrometry requires a very homogeneous magnetic field (shimming), a volume voxel and a sufficient number of measurements
  • Principal metabolites studied in MRS of the 1H nucleus: see tab. 15.1
  • Any method of spectroscopy, calls for suppression of the water signal (CHESS), and possibly of the fat signal: present in large quantities in the body, these have a masking effect on the metabolites close to their resonance peaks.
  • The basic spectroscopy sequences are PRESS and STEAM. PRESS records a spin echo whereas STEAM only records a stimulated echo, of weaker intensity. Both have an excitation pattern comprised of 3 RF pulses.
  • In single voxel spectroscopy the 3 RF pulses select the voxel of interest, located at the intersection of the 3 orthogonal planes. The recorded echo comes from the voxel submitted to the 3 RF pulses only.
  • In chemical shift imaging (CSI), the 3 RF pulses select a slice or volume that is spatially encoded by phase gradients. There are different methods of accelerating CSI data acquisition. Chemical shift imaging yields multiple spectra of the slice or volume of interest. These are represented as a parametric image or studied separately.
  • Quantification tends to be relative, although it is possible under certain conditions, after calibration, to produce absolute quantification of the metabolite concentration.
  • The fields of application of magnetic resonance spectrometry are essentially those of: tumors, inflammatory and infectious pathologies and metabolic pathologies, mainly of the brain. Numerous developments are taking place in regard to other organs (prostate, breast, bones and joints…)

References

  1. Galanaud, Nicoli. . Journal de radiologie. 2007 Mar;88(3 Pt 2):483-96.
  2. Jansen, Backes. 1H MR spectroscopy of the brain: absolute quantification of metabolites. Radiology. 2006 Aug;240(2):318-32.
  3. Burtscher and Holtas. Proton MR spectroscopy in clinical routine. J Magn Reson Imaging. 2001 Apr;13(4):560-7.
  4. Dydak and Schar. MR spectroscopy and spectroscopic imaging: comparing 3.0 T versus 1.5 T. Neuroimaging clinics of North America. 2006 May;16(2):269-83, x.
  5. Mullins. MR spectroscopy: truly molecular imaging; past, present and future. Neuroimaging clinics of North America. 2006 Nov;16(4):605-18, viii.
  6. Kwock. Localized MR spectroscopy: basic principles. Neuroimaging clinics of North America. 1998 Nov;8(4):713-31.
  7. Pohmann, von Kienlin. Theoretical evaluation and comparison of fast chemical shift imaging methods. J Magn Reson. 1997 Dec;129(2):145-60.
  8. Bartella, Morris. Proton MR spectroscopy with choline peak as malignancy marker improves positive predictive value for breast cancer diagnosis: preliminary study. Radiology. 2006 Jun;239(3):686-92.
  9. Katz and Rosen. MR imaging and MR spectroscopy in prostate cancer management. Radiologic clinics of North America. 2006 Sep;44(5):723-34, viii.

Magnetic Resonance Spectroscopy (MRS)

Magnetic resonance spectroscopy allows noninvasive and in vivo exploration of the molecular composition of tissue. It identifies certain molecular constituents - the metabolites - involved in physiological or pathological processes. Even though spectroscopy can be performed on different nuclei, we will focus here on the spectroscopy of the hydrogen nucleus, by far the most widely studied in clinical MRI.


Chemical shift, spin-spin interaction and J-coupling

advertisementChemical shift corresponds to a change in the resonance frequency of the nuclei within the molecules, in function of their chemical bonds. The presence of an electron cloud constitutes an electronic shield that slightly lowers the B0 magnetic field to which the nucleus would normally be subjected. The resonance frequency difference is expressed as parts per million or ppm, a value that is independent of the amplitude of the magnetic field. The value of the chemical shift thus provides information about the molecular group carrying the hydrogen nuclei.

Chemical shift value

For a given molecule, the chemical shift in ppm versus the reference molecule is defined by the relationship:
With:
  • ωm: resonance frequency of the studied molecule
  • ωref: resonance frequency of the reference molecule
  • dm: chemical shift value in ppm
In chemical shift, the difference in frequency compared to the reference molecule is proportional to the amplitude of the B0 magnetic field.
Interaction between the atomic nuclei of neighboring chemical groups translates as each peak breaking down into a complex peak (doublet, triplet, multiplet): this is spin-spin coupling. The spacing between these peaks has a fixed frequency value (Hz), called the J-coupling constant, independent of magnetic field amplitude.
A spectrum is represented:
  • as an abscissa: metabolite position according to chemical shift. Tetramethylsilane is used as the axis origin reference, a molecule used in in-vitro NMR, which is conventionally at 0 ppm. This axis is orientated from right to left.
  • as an ordinate: peak amplitude.
MRS spectrum
The area below the peak curve will basically be determined by metabolite concentration. The width of the peak is inversely proportionate to T2* relaxation time.

Equipment and software required for MRS

In-vivo MRS uses MRI apparatus that is virtually identical to that used in imaging, but nevertheless requires:
  • A sufficiently strong and very homogenous magnetic field, to distinguish resonance peaks (at least 1.5 T, shimming)
  • Specific sequences for spectroscopic signal acquisition. There are two types of MRS: either SVS: Single Voxel Spectroscopy which receives the spectrum from a single voxel only, or spectroscopic imaging (CSI: Chemical Shift Imaging) which measures spectra in projection (1D), on a slice (2D) or a volume (3D).
  • Adapted data processing software
  • An adapted radiofrequency sustem (in the resonance frequency of the studied nucleus).

Field homogeneity, SNR and spectrum quality

Analysis of the differences in metabolite resonance frequency can only be performed in the presence of a highly homogenous magnetic field. A heterogeneous magnetic field leads to resonance frequency dispersion, spreading out the peaks or even causing them to disappear into the background noise.
Prior to any MRS acquisition, the magnetic field is homogenized (shimming) in the region of interest. The bigger the region, the harder it is to homogenize the magnetic field throughout. Close to the bone, calcifications or hemorrhagic zones, spectroscopic quality will be poorer due to perturbations in the field generated by the differences in magnetic susceptibility compared to soft tissue. The precession frequency of the water must be optimized to adequately suppress the water peak, using selective frequency pulses and dephasing gradients.
The other problem with MRS concerns the weak signal-to-noise ratio. This entails multiplying the number of measurements (NSA) and limits spatial resolution (voxel of a minimum volume of roughly 3.5 cm3 i.e. of dimensions 1.5 x 1.5 x 1.5 cm).

Spectrum quality is evaluated according to two main criteria:
  • signal-to-noise ratio (height of metabolite peaks in relation to background noise)
  • spectral resolution (peak width, which determines whether the different metabolites can be separated).
Spectral resolution will depend on the homogeneity of the magnetic field B0 and on digital resolution, i.e. the precision with which the signal is sampled, determined according to sampling time (Te = 1/Fe) and the total number of points measured.

Metabolites explored in 1H-MRS

Metabolites explored and TE

Only a limited number of molecules with protons are observable in MRS.
In brain MRS, the principle molecules that can be analyzed are (fig. 15.1):
  • N-acetyl-asparate (NAA) (molecule present in healthy neurones) at 2.0 ppm
  • Creatine/phosphocreatine (Cr) (energy metabolism molecules) at 3.0 ppm
  • Choline compounds (Cho) (marker in the synthesis and breakdown of cell membranes) at 3.2 ppm
  • Myo-inositol (mI) (only found in glial tissue) at 3.5 ppm (figure 15.2)
  • Glutamine-Glutamate-GABA complex (Glx) (neurotransmitters) between 2.1 and 2.5 ppm
  • Lactate (Lac) (anaerobic metabolism): doublet at 1.35 ppm (figure 15.3)
  • Free lipids (Lip): wide resonance, doublet at 1.3 and 0.9 ppm
In prostatic MRS, the citrate peak is also looked for at 2.6 ppm.
The number of discernable metabolites will vary according to the TE of the spectroscopy sequence: the longer the TE (135 or 270 ms), the more long T2 metabolites are selected. With short TE (15 to 20 ms), the spectrum will be more complex because of the greater number of superimposed peaks, producing a number of problems for quantification and interpretation.

Principal metabolites

Metabolite Freq. (ppm) short
T2
long
T2
Role Anomalies
mI myoInositol 3,6 Glial marker ↑ : gliomas, MS reactional gliosis
↓ : herpetic encephalitis
Cho Choline 3,2 Cell membrane metabolism marker ↑ : tumors, demyelinization
Cr
Pcr
Creatine
Phosphocreatine
3,0 Energy metabolism marker, serves as reference peak as it is ~ constant
Glx GABA, Glutamate
Glutamine
2,1-2,5 Intracellular neurotransmitter marker ↑ : hepatic encephalopathy
NAA N-Acetyl-Aspartate 2,0 Healthy neuron marker ↑ : Canavan's disease
↓ : neuronal distress
Succ Succinate 2,4 D
I
S
E
A
S
E
S
Pyogenic abscess
Ac Acetate 1,9 Abscess
Ala Alanine 1,5 (doublet) Meningioma, Abscess
Lac Lactate 1,3
(doublet)
● + ● – Ischemia, convulsions, tumors, mitochondrial cytopathies ↑ : anaerobic metabolism
Lip Free lipids 0,9
1,4
Necrotic tumor (high grade)
aa Aminoacids 0,97 Pyogenic abscess

Single voxel spectroscopy (SVS)

 In SVS, the signal is received of a volume limited to a single voxel. This acquisition is fairly fast (1 to 3 minutes) and a spectrum is easily obtained. It is performed in three steps:
  • Suppression of the water signal: the quantity of hydrogen nuclei in the water molecules in the human body is such that the water peak at 4.7 ppm “drowns” and masks the spectroscopic signal from the other metabolites. It is therefore vital to suppress the water peak to observe the metabolites of interest.
  • Selection of the voxel of interest
  • Acquisition of the spectrum, for which two types of sequence are available (PRESS: Point-RESolved Spectroscopy, STEAM: STimulated Echo Acquisition Mode)

Water signal suppresion

The most commonly used method to suppress the water peak is CHESS (CHEmical Shift Selective). CHESS consists in applying three couples (90° RF pulses + dephasing gradients) in each spatial direction. The bandwidth of these RF pulses is narrow and centered on the resonance frequency of the water peak in order to saturate the water signal and preserve the signal from the other metabolites.

Techniques applying a 180° inversion pulse with adapted TI, like those used in FLAIR and STIR sequences, can also be used to eliminate the water signal (WEFT: Water Elimination Fourier Transform) or suppress the fat signal in breast spectroscopy, for example. In practice, CHESS is more commonly used than WEFT.

Principles of volume selection

The analyzed volume is selected by a succession of three selective radiofrequency pulses (accompanied by gradients) in the three directions in space. These pulses determine three orthogonal planes whose intersection corresponds to the volume studied. Only the signal of this voxel will be recorded, by selecting only the echo resulting from the series of three radiofrequency pulses.
Single Voxel Selection
Single Voxel Selection

PRESS and STEAM sequences

Acquisition of the signal from the selected voxel can be performed using two different types of sequence:

STEAM (Stimulated Echo Acquisition Mode)

 The three voxel-selection RF pulses have flip angles of 90°. The stimulated echo is recorded from the cumulated effect of the three pulses, thus corresponding to the signal from the only voxel of interest (cf. chapter 6.7.2. Hahn echo and stimulated echo). The TE of the stimulated echo corresponds to double the time interval between the first two pulses. The delay between the second and third RF pulses is the mix time TM. This technique is particularly adapted to short TE spectral acquisitions.

PRESS (Point RESolved Spectroscopy)

 In the PRESS method, the RF pulses have flip angles of 90° - 180° - 180°. The signal emitted by the voxel of interest is thus a spin echo. The amplitude of this spin echo is two times greater than the stimulated echo obtained by STEAM. The PRESS technique thus offers a better signal-to-noise ratio than STEAM. It can be used with short TE (15 – 20 ms) or long TE (135 – 270 ms).
Whatever the case spectroscopic signal recording does not use a frequency encoding readout gradient, as the frequency is used to constitute the spectrum (rather than the position), after Fourier transform of the signal.

Spectroscopic imaging (CSI : Chemical Shift Imaging)

Metabolic imaging (CSI) consists in recording the spectroscopic data for a group of voxels, in slice(s) (2D) or by volume (3D). It is based on a repetition of STEAM or PRESS type sequences to which is added spatial phase encoding. The number and direction of phase encodings depend on the number of dimensions explored (1D, 2D or 3D), adding on to acquisition time. The duration of the sequence is equal to TR ∙ Nph1D ∙ Nph2D ∙ Nph3D ∙ NSA (NphxD number of phase encoding steps in direction x).
To reduce acquisition time, variants of the basic sequence have been proposed:
  • Multiple Slice CSI accelerates the acquisition of several slices compared to 2D CSI
  • Turbo CSI (several echoes received by rewinding the phase encoding gradient before each additional echo, derived from the repetition of the last spin echo pulse and the gradient)
  • Fast CSI provides an important speed gain compared to 2D CSI, by performing spatial encoding in one direction during signal acquisition, by means of oscillating gradients, similar to spatial encoding in an echo planar sequence. These techniques are less sensitive than the classic CSI sequence. The speed gain is particularly useful in 3D spectroscopic imaging or in the case of motion artifacts.
  • CSI with parallel acquisition (SENSE CSI).
Signal processing calls on Fourier transforms (1 for each phase-encoded dimension phase + 1 for spectral analysis) and requires data correction (correction of the baseline, phase, smoothing truncation artifacts…).
The results appear in the form of parametric images («metabolic maps ») or a matrix of the spectra of the regions to be studied

Signal processing in MRS

Extracting a quality spectrum from the signal involves several processing steps:
  • Zero-filling or apodization filtering to complete the digitized FID signal
  • Phase correction to obtain the real part of the spectrum (absorption spectrum)
  • Baseline correction

In vivo metabolite quantification

  • Relative quantification: The results of the MRS are generally expressed as concentration ratios. Creatine peak or comparison with the healthy controlateral zone often serve as the reference values.
  • Absolute quantification: Measuring the true concentration of metabolites by MRS comes up against several technical difficulties: the peak area has to be determined accurately then converted into concentration after calibration.

Main clinical applications of MRS

With greater feasibility in the clinical practice setting, the indications for MRS are multiplying. In brain MRS, it yields diagnostic data for:
  • Tumoral pathologies: choline, myo-inositol (gliomas), free lipids (necrotic tumor: glioblastoma, metastasis…), alanine (meningioma), lactate
  • Demyelinizing inflammatory pathologies: myo-inositol,
  • Infectious diseases: free acetate and aminoacids in certain abscesses, reduction of NAA and VIH encephalopathy
  • Metabolic pathologies: myo-inositol and glutamine / glutamate in hepatic encephalopathy, lactate and diffuse cerebral distress
  • Epilepsy, brain maturation and degenerative diseases…

No comments:

Post a Comment