In vivo MRS allows for the noninvasive measurement of the levels of some compounds in body tissues. It exploits the magnetic properties of certain atomic nuclei that are present in these molecules. The nuclei that are best accessible for in vivo MRS experiments are those of proton (1H), phosphorus (31P) and carbon-13 (13C) atoms.
The following sections present a short introduction of MR spectroscopy to provide the reader with sufficient background to understand the biological and clinical applications of MR spectroscopy. For more in-depth information the reader is referred to other publications, see for example[6
The key feature of MR spectroscopy is that certain biochemical compounds, mostly metabolites, can be identified in an MR spectrum by their specific spectral pattern, which is composed of one or more distinct signals. The intensity of the signal is proportional to the tissue amount of a certain nuclei and thus reflects the tissue levels of the compound in which it is present. An example of a typical in vivo 1H MR spectrum of a healthy liver is shown in Figure . The horizontal axis of a spectrum represents the resonance frequency or chemical shift (both terms are explained below), the vertical axis represents the signal intensity. This spectrum is dominated by three main signals and in addition there are some smaller peaks, and broad underlying resonances. In the next few sections we explain in more detail how this spectrum is obtained and what particular information can be extracted from it.
Figure 1 In vivo 1H magnetic resonance spectra of human liver tissue obtained from a healthy volunteer on a 3.0T magnetic resonance system. Above: Spectrum with unsuppressed water signal; Below: Spectrum with partial suppressed water signal, showing the overlapping (more ...)
Atomic nuclei with unpaired neutrons and/or protons, are detectable by nuclear magnetic resonance (NMR). As the nucleus is spinning around its axes and bears an electric charge it is associated with a magnetic dipole that can be seen as a tiny bar magnet (Figure ). Outside a magnetic field these nuclear spins or tiny magnets have a random orientation; however, when placed in a strong constant external magnetic field B0 they will become aligned (Figure ). The nuclear spins of the atoms 1H, 31P and 13C can be oriented parallel or anti-parallel to B0. However, the spins do not exactly align but are at an angle to B0. This causes them to precess around the axis of B0 with the so-called Larmor frequency ν0 = ω0/2π = γB0/2π where the gyromagnetic ratio γ has a specific value for each nucleus (Figure ). This implies that every type of nucleus has a different precession frequency, proportional to the B0 field strength. At a field strength of 3T, which is commonly used for human applications, these frequencies for 1H, 31P and 13C are: 127.7 MHz, 51.8 MHz and 32.1 MHz respectively, which is in the radiofrequency range.
A 1H nucleus spinning around its axes (left) can be regarded as a tiny bar magnet (right).
Outside a magnetic field spins have a random orientation (left) but spins inside a strong constant magnetic field B0 will become aligned (right).
A spin precessing with the Larmor frequency around the axis of B0 at a small angle with respect to this axis.
The parallel and anti-parallel orientations are associated with a low and a high energy state respectively. The energy difference between the two spin states equals
E = h γ B0 [Equation 1],
where h is Planck’s constant (h = 6.626 × 10-34 J/s). At room temperature, there are slightly more spins in the lower energy level, Nα, than in the upper level, Nβ. The distribution over these energy levels is given by Boltzmann statistics
Nβ/Nα = e-E/kT [Equation 2],
where k is Boltzmann’s constant (1.3805 × 10-23 J/K) and T is the temperature in Kelvin. The population difference results in a net macroscopic magnetization M0. This so-called longitudinal magnetization is aligned parallel to B0 (Figure ). Only the net magnetization is detectable and its extent determines the achievable signal-to-noise ratio (SNR). At 1.5T and 37 °C (310 K), the population difference represents only a small fraction (about 10-6) of the total spin population, which explains why MR is a relatively insensitive technique. It follows from Equations 1 and 2 that sensitivity can be improved with a higher magnetic field B0 or a lower temperature.
Figure 5 The direction of the main magnetic field B0, is commonly placed along the z-axis and the magnetization along this axis is Mz, which at equilibrium, equals M0. Left: Multiple individual spins precessing with the Larmor frequency around the axis of B0. (more ...)
Flip angle, free-induction decay, T1 and T2
The direction of the main magnetic field B0, is commonly placed along the z-axis and the magnetization along this axis is Mz, which at equilibrium, equals M0 (Figure ). The longitudinal magnetization however is not detectable as it is “overruled” by the main magnetic field B0. To detect the net macroscopic magnetization M0, an radio frequency (RF) pulse with a magnetic field perpendicular to the main magnetic field B0 and with a frequency equal to the precession frequency is sent with an RF transmitter coil. As a consequence of the applied RF pulse M0 magnetization will rotate away from the z-axis toward the transverse plane (x-y plane). The angle to which the net magnetization is rotated relative to the main magnetic field direction is called the flip angle. A so called 90° RF excitation pulse will therefore rotate M0 into the transverse plane. The spin population of the energy levels becomes equal and spins precess coherently (Figure ). However, after the RF excitation pulse the spins start to return towards the original energy level distribution and towards incoherent precession (dephasing) and after a while the system will be in equilibrium again. The time constant which describes how the magnetization returns to the original longitudinal alignment is called the spin lattice relaxation time T1: Mz = M0(1-e-t/T1). The time constant which describes the return to incoherent precession is called the spin-spin relaxation time T2: Mxy = Mxy0 e-t/T2. By definition T1 is longer than T2.
Figure 6 Multiple spins precessing with the Larmor frequency around the direction of B0 (left). After a 90° radio frequency (RF) pulse, the spin population of the energy levels become equal and the spins precess coherently (middle). The magnetization vector (more ...)
After the RF pulse, the transverse component of the M0 magnetization precesses with the Larmor frequency at resonance and will induce a current in the RF coil which is now switched to receive mode. The decaying signal of this component is called the free-induction decay (FID) response signal. This digitally recorded FID signal is mathematically converted by a Fourier transform from the time to the frequency domain, which results in a so called MR spectrum, that may contain one or more resonances, signals or peaks at particular frequencies.
Shielding and chemical shift
An MR spectrum of the liver would not be very interesting if all nuclei of a certain type resonate at the same frequency. Fortunately, each nucleus in a given molecule is shielded from the main field by a weak opposing field from the surrounding electrons, induced by and also proportional to B0. The amount of shielding by these electrons highly depends on the chemical environment of the nucleus. This shielding, which is generally unique for each chemically distinguishable site in a molecule, is expressed as the shielding constant σ, and the total effective field experienced by a given nucleus is: Beff = B0 (1-σ). The resulting change in resonance frequency νeff = ωeff/2π = γ B0 (1-σ)/2π relative to that of a chosen reference compound, νref, is generally referred to as the chemical shift σ = (νeff - νref )/νref expressed in units of ppm (1 ppm = 100 Hz at ν0 = 100 MHz). Thus the chemical shift is the key property of MR spectroscopy which enables detection of a wide variety of chemical groups and metabolites containing these groups. Special RF pulses are used to excite a band of frequencies covering the chemical shift range of a particular nucleus in a biological sample.