1
MRI Principles, Hardware Components and Quantification

Hervé SAINT-JALMES1, Hélène RATINEY2 and Olivier BEUF2

1 LTSI, Inserm, Université de Rennes, France

2 CREATIS, CNRS, Inserm, INSA Lyon, Université Claude Bernard Lyon 1, France

1.1. Introduction

The demonstration of the nuclear magnetic resonance (NMR) phenomenon in condensed matter in 1946 (Bloch 1946; Purcell et al. 1946) certainly stems from significant progress accrued during the previous decade and in particular during the war with the development of radars and radiofrequency technology. Today, a similar consideration can be made for magnetic resonance imaging (MRI), where technological progress contributes significantly to the improvement of measurement instrumentation and therefore to the quality of signal and images produced by these devices. Such progress paves the way for imaging techniques that are not exclusively qualitative but also quantitative.

This introductory chapter concisely describes MRI principles and the main components of an MRI scanner, in particular those responsible for the generation of magnetic fields and signal acquisition.

Schematically, several steps are required to obtain a localized spectrum or an image (see Figure 1.1):

1. the system of spins in a sample must reach thermal equilibrium, which leads to a macroscopic magnetization M0 being produced in the presence of a static polarization field B0;
2. the equilibrium of the macroscopic magnetization is perturbed by tilting the M0 vector away from its equilibrium position by means of a radiofrequency magnetic field B1;
3. spatial encoding is performed using magnetic field gradients;
4. the system’s response following the perturbation is observed by gathering the MR signal by means of a radiofrequency coil;
5. the spectrum or image is reconstructed and quantitative parameters are extracted.

In the following, we will focus on recalling the physical principles of MRI along with the associated technological solutions and their implementation (Saint-Jalmes 2022) up to a general overview of quantitative MRI approaches that are amply presented in this book.

Figure 1.1. Sequence of four key steps of an MRI scan of an organ, with a fifth step of image/spectrum reconstruction and extraction of quantified parameters

In NMR and MRI, the denomination of magnetic field B is commonly used. From a physical point of view, B is a fundamental quantity, similar to the electric field E in electrostatics (Gié et al. 1976). In the past, the mathematical analogy between vectors E and H led to H being called the magnetic field, with B being then referred to as magnetic induction. The latter term will not be used in this work, although it is still widely used.

1.2. Macroscopic magnetization and static magnetic field B0

1.2.1. Nuclear magnetization

NMR makes use of the magnetic properties of nuclei having a non-zero spin hI. Among nuclei with such a feature, the most commonly studied are the hydrogen nucleus (1H or proton), phosphorus-31 (31P), sodium-23 (23Na), fluorine-19 (19F) and carbon-13 (13C) (see Chapter 9). In a nucleus with a spin number I (integer or half-integer), the magnetic nuclear moment μ is aligned with the spin moment, such that μ = γħI, where γ, called the gyromagnetic ratio, is a characteristic constant of the nucleus of interest and ħ is the reduced Planck’s constant h/2π with h = 6.62 × 10−34 J.s.

Whenever a sample with a large number N of identical nuclei having spin number I is subjected to a magnetic field B0, it possesses a resulting magnetization vector M0 according to the Boltzmann equation:

where kB is the Boltzmann constant (kB = 1.38 × 10−23 J.K-1) and T the absolute temperature. At thermal equilibrium, the macroscopic magnetization M0 has only one longitudinal component oriented along B0, which can be expressed as M0 = χ0 B0, where χ0 is the nuclear magnetic susceptibility. This nuclear magnetization, the evolution of which is inversely proportional to temperature T, is extremely small (typically of 4.8 pA/m for 1 mm3 of water at 1.5 T and T = 300 K) (Abragam 1961).

1.2.2. Magnet

An MRI system is similar to a chain: the weakest component determines the final performance of the entire system. In this regard, a significant role is played by the magnet as its choice and features directly influence image quality together with the purchase and operations costs of an MRI device.

1.2.3. Roles and orders of magnitude

The static magnetic field is employed both for polarizing the sample and imposing the frequency of excitation and signal acquisition. These two aspects determine the specifications and constraints of the magnet. In the remainder of this chapter, we shall limit ourselves to the description of devices for clinical imaging; nevertheless, most of the arguments are largely transposable to preclinical imaging (i.e. imaging used in animal model studies).

1.2.3.1. Field strength

Performing medical MRI scans at fields lower than a tenth of a tesla seems difficult since the field should be sufficiently high in order to produce an image with an acceptable spatial resolution within a reasonable acquisition time. Indeed, the lower the field, the lower the signal is, and therefore the signal-to-noise ratio (S/N) is low.

By increasing the field strength, S/N is increased and therefore the spatial resolution and/or the scan time is reduced (a trade-off between the two always exists). However, the higher the field, the more challenges arise related to the propagation of radiofrequency waves and safety aspects (see Chapter 12). Furthermore, the cost of a magnet and that of the entire MRI scanner increase significantly with field strength. Thus, magnets with fields higher than 3 T are seldom encountered in clinical routine, with ultra-high-field magnets nowadays typically built to satisfy research needs. Finally, regulations (2013/35/UE) describe the minimal safety indications for the staff exposed to magnetic fields.

1.2.3.2. Volume of interest, spatial uniformity and temporal stability

The strong magnetic field (between a few tenths of a tesla and a few teslas) has to be induced over a volume of interest (e.g. tens of liters in clinical imaging) while allowing the scanned body to access the center of the magnetic field region.

In order to produce a spatially well-resolved image or a fine spectrum, the magnet’s B0 field must be as uniform (or homogeneous) as possible over the volume of interest. For whole-body imaging, uniformity of the order of a part per million over volumes of several tens of liters is sought. As an example, in a field of 1.5 T, 1 ppm corresponds to a tiny field variation of 1.5 μT. At the same time, extremely high stability of the field, of the order of 0.1 ppm, is expected over the duration of an exam (between 30 min and 1 h).

1.2.4. Technical approaches

To create such a magnetic field, several approaches are available. One of these consists of using the magnetic energy stored in a permanently magnetized material, a magnet. Magnetic fields of the order of tenths of a tesla can be created using ferrites or NdFeB. The shape of the magnet and/or polar ferromagnetic parts can channel the magnetic flux for a uniform distribution inside the region of interest. The undeniable advantage of this approach is the absence of energy sources. On the other hand, the magnet’s mass is considerable (e.g. about ten tons for a field of 0.3 T). In addition, the field intensity in such magnets varies noticeably with temperature (typically 0.1 %/°C). The race toward higher fields has certainly made this approach marginal, although it offers several advantages.

Figure 1.2. Magnet with two pairs of coils (axisymmetric around the z-axis) and representation of the magnetic fields created along the symmetry axis of the magnet. The two red coils create a strong field at the center of the magnet, together with a significant second-order component (red curve). When the blue pair of coils are added, despite being less efficient, they produce a negative second-order component (blue curve) which can compensate for the imperfection and provide a uniform field (green curve)

The most widely employed approach consists of a constant direct current flowing in a conducting or superconducting material. The superconducting approach is unavoidable for creating strong fields used for clinical imaging.

The design of the first MRI magnets was based on the pioneering work by Garrett in the 1950s. Garrett was the first person to solve the magnet design problem (Garrett 1951) by expressing the field distribution in the form of zonal harmonics. For this series, the goal consists of canceling as many components as possible up to the 12th, 16th or even 20th order via...