Basic physics

MRI is based on nuclear magnetic resonance, the phenomenon of the resonance of atomic nuclei in response to radiofrequency (RF) waves.

The hydrogen atom is the simplest and most abundant element in the body and consists of one proton nucleus orbited by one electron. The hydrogen nucleus can therefore also be termed a proton, and current clinical MRI techniques are based on receiving and processing RF signals from protons. Protons have a magnetic axis which is normally randomly orientated. When a magnetic field is applied, the protons align in synchrony and spin around an axis in line with the main magnetic field—this spinning is termed precession. The rate at which protons precess is measured by the precession frequency, which changes linearly with increasing magnetic field strengths. When protons precess in synchrony they are said to be in-phase. There is loss of synchrony with time, and this is also termed out-of-phase.

At equilibrium within a magnetic field, overall proton alignment is in the direction of the main magnetic field and they have net longitudinal magnetization. This equilibrium can be disturbed by transmission of RF energy at the precession frequency of the proton which is 63 megaHertz (MHz) for water protons at 1.5 Tesla (T)—the strength of most commercially used magnets (Figure 1.1).

Figure 1.1 1.5T CMR Scanner. The adjustable patient table is shown within the surrounding superconducting magnet.

The degree of proton excitation is proportional to the amplitude and duration of the RF pulse. After excitation, proton relaxation occurs as the energy is dissipated and this process is defined by two parameters known as T1 and T2. T1 relaxation times measure the time after excitation to recover the longitudinal magnetization found in the equilibrium state. Transverse magnetization decays at a rate measured by T2, which is faster than the rate of T1 recovery. T1 and T2 relaxation vary according to the environment of the hydrogen atom within tissues and imaging sequences can be designed with different preference (or weighting) to one of these relaxation parameters for tissue characterization, known as T1-weighted (T1W) and T2-weighted (T2W) acquisitions. The values for T2 are always below that of T1, and T1 represents the upper limit of T2. T1 and T2 values tend to parallel each other when proton motion is relatively random, for example in adipose tissue, which has a short T1 and T2, and free water, which has a long T1 and T2. Tissues with a more organized structure contain abundant bound water. In this case proton motion is not random, there is increased transverse decay from the exchange of energy between protons, and T2 values become shorter than those of T1.

Localization of anatomical position within a selected imaging slice or volume is done with the application of frequency- and phase-encoding gradients. The corresponding direction of application of these gradients is known as the frequency encode or phase encode direction. With modifications of the phase-encoding gradients flowing blood can be differentiated from stationary anatomy via alterations in the phase of the MR signal. The velocity of material is proportional to the phase change or phase shift caused by its movement during gradient application.

Transmission and reception of RF energy is via special aerials known as coils with subsequent conversion of these raw data into images using ultrafast computers and a process known as Fourier transformation.

Main sequences

There are two fundamental types of sequence commonly used in CMR: gradient echo (GE) and spin echo (SE). As a general rule, with GE sequences both blood and fat appear white and so this technique is also known as white-blood imaging (Figure 1.2a). By contrast, in SE sequences blood is usually black but fat is white, giving rise to the term black-blood imaging (Figure 1.2b). SE sequences are more useful for anatomical imaging as opposed to the functional imaging performed with GE sequences. Variations of GE sequences are fast low-angle shot (FLASH), fast imaging with steady-state free precession (SSFP), and velocity mapping. GE imaging also forms the basis of the inversion recovery technique.

Figure 1.2 (a) CMR of mid-ventricular short-axis view using GE imaging. Blood within the left ventricle (LV) is white as is the surrounding epicardial fat (black arrow). The LV myocardium (white arrow) appears mid-grey. (b) CMR in the same plane using SE imaging. Blood within the LV now appears black while the fat remains white.

Areas of focal myocardial dysfunction and abnormal flow patterns are readily visualized with the technique of cine imaging using SSFP (or cines). Cines are obtained by rapid repetition of a variant of the basic GE sequence to obtain a series of cardiac images at progressively advancing points of the cardiac cycle which when put together form a cine loop. The weighting of SSFP sequences depends on the ratio of T2/T1, therefore most fluids and fat have a high signal and appear white. However, muscle and many other solid tissues have a long value for T1 and a short value for T2. This means that their signal intensity is reduced to shades of grey (Figure 1.2a). In addition to cines, the SSFP sequence can also be applied as a two-dimensional (2D) single-shot technique, as a real-time technique (not requiring breath holding or electrocardiogram (ECG) triggering), and as a three-dimensional (3D) volume scan.

Velocity mapping (or flow velocity mapping) techniques can determine the average velocity within a single imaging voxel, typically 1 ×1 ×10 mm³. The operator selects the required plane and sets a maximal encoding velocity (Venc; Figure 1.3). The initial Venc used is an approximation of the velocity expected based on factors such as clinical history, the type of valve or conduit lesion, and images already acquired. The Venc represents the practical upper limit of velocities that can be depicted unambiguously and should ideally be set to a numerical value just greater than the true velocity. Problems occur if it is set much higher or lower than this value—with the former leading to less sensitivity and the latter causing misrepresentation via the phenomenon of aliasing. Velocity aliasing is when the flow appears to be in the opposite direction and is characterized by a sudden transition of white-toblack or vice versa within the chosen flow region (Figure 1.3d). To eliminate or reduce aliasing a higher Venc must be set and the velocity mapping sequence repeated (Figure 1.3e). The maximal velocity of jets under interrogation should only be determined from the Venc-optimized images. Aliasing is also noted with 2D Doppler echocardiography and the technique of velocity mapping gives similar information. Subjects are asked to suspend their breathing for measurement of peak velocities using this method, which takes approximately 20 seconds to perform. An example of its use is for the quantification of peak velocity in valvular stenosis. Velocity mapping sequences are also used to calculate overall flow in a major vessel through the cardiac cycle and so can be used for quantification of regurgitation in valvular incompetence. For calculation of peak velocity and transvalvular flow, the plane used must be through plane—a plane perpendicular and mjust distal to the area of interest. Velocity mapping CMR is also used to confirm abnormal chamber communication and the ratio of pulmonary to systemic flow in shunts such as septal defects.

Figure 1.3 Cine images (a, b) combined with velocity mapping imaging (c–e). (a) Left ventricular outflow tract (LVOT) cine image of severe aortic stenosis (AS). The plane chosen for velocity mapping is just distal and perpendicular to the valve orifice (white line) and results in the oblique sagittal view shown in (b). (b) This shows a tricuspid aortic valve (AV; white arrow) and is the plane used for subsequent velocity mapping. AV area can be measured in this view. (c) Magnitude image component of the velocity mapping sequence, which confirms reduced valve area in a tricuspid AV. (d) The velocity mapping sequence shows systolic blood flow through the AV as black, but the Venc was set too low by the operator and the centre of the jet has turned white, a phenomenon known as aliasing (white arrow). (e) A second velocity mapping sequence was therefore performed using an increased Venc. The black jet through the AV is now clearly seen (white arrow) and there are only occasional edge pixels which are aliased. Peak systolic velocity in the centre of the stenotic jet was 4 m/s.

Compared to GE sequences, SE pulse sequences are usually more robust to system imperfections, such as magnetic field inhomogeneities. In SE imaging, structures need to be stationary for the delivery of two RF pulses. The T2 value of stationary fluid is long and gives high signal. Examples are fluid-filled structures such as cysts, which therefore appear white with SE sequences. However, flowing blood moves out of the selected slice before receiving the second pulse and so gives no signal and appears black (Figure 1.2b). Slower flowing blood can give persistent signal of varying signal intensity. Important variations of SE sequences are fast (or turbo) spin echo (FSE or TSE). FSE allows faster imaging than standard SE by acquiring more lines of data for every RF pulse delivered and allows acquisition of an entire image in a single heartbeat.

The inversion recovery technique uses a prepulse to create high T1 tissue contrast which is important for infarct imaging. This sequence and contrast-enhanced magnetic resonance angiography (CE-MRA) require use of a contrast agent. MRI contrast agents are commonly based on chelates of gadolinium which are paramagnetic, one example being gadolinium diethylenetriamine pentaacetic acid (Gd-DTPA). All gadolinium chelates currently approved for clinical use are extravascular and therefore become distributed within the interstitium following initial intravenous delivery.

CMR is performed by applying these main sequences and their variants to evaluate cardiovascular physiology and anatomy, characterize tissue, and perform vascular angiography. Additionally, cardiac metabolism can be determined with magnetic resonance spectroscopy (MRS). MRS is not covered in this book and the interested reader is referred to the further reading section.

Most CMR scans are timed with respect to the ECG (ECG-gated) to minimize cardiac motion artefact, and subjects are asked to suspend their breathing in end expiration (breath-hold) to minimize respiratory motion artefact. In some cases scans can be linked to the respiratory cycle using diaphragmatic monitoring techniques (respiratory-gated), allowing subjects to breathe normally (free breathing). High signal from fat can be reduced by the application of a frequency-selective prepulse (fat suppression).

Patients are advised that data are being acquired during the time when the scanner is making a noise. This noise is generated by coils within the magnet. Headphones are worn by subjects during a CMR study and this serves to minimize their discomfort during noisy periods, facilitate hearing instructions from scanner operators, and allow music to be heard throughout if requested.

Hardware and software requirements

CMR is performed within a large superconducting magnet with its associated radio wave generation and computer systems. Medical gases, physiological monitoring telemetry, stress infusion pumps for delivery of adenosine and dobutamine, a power injector for contrast studies, and full resuscitation equipment and drugs are also required. As mentioned above, microphones, headphones and a music system are important. Ear plugs are needed for people who are required to accompany certain patient groups such as children, and those with varying degrees of claustrophobia. Less than 2% of scans are limited by claustrophobia in the adult population and other measures of alleviating anxiety are blindfolds and prism glasses. The latter provide a way for the patient to look out even when enclosed by the bore of the magnet.