Principles of magnetic resonance imaging

INTRODUCTION — Magnetic resonance imaging (MRI) is an important tool in the diagnosis and evaluation of diseases [1]. In the early 1970s, Paul Lauterbur and Raymond Damadian applied nuclear magnetic resonance (NMR) technology to the imaging of living organisms, generating images referred to as zeugmatographs [2-5]. Subsequent refinements in image acquisition and processing, developed by Sir Peter Mansfield and others, allowed improved visualization of anatomic detail and broader clinical application of MRI [1,6-8]. Lauterbur and Mansfield were awarded the 2003 Nobel Prize in Medicine and Physiology for their contributions to medical imaging.

This topic will review the principles of MRI. Clinical applications of MRI are discussed in individual topic reviews, including:

●(See "The role of magnetic resonance imaging in prostate cancer".)

●(See "Magnetic resonance imaging changes related to acute seizure activity".)

●(See "MRI of the breast and emerging technologies".)

●(See "Cardiac imaging with computed tomography and magnetic resonance in the adult".)

●(See "Clinical utility of cardiovascular magnetic resonance imaging".)

●(See "Magnetic resonance imaging of the thorax".)

MAGNETIC RESONANCE PHYSICS — The phenomenon of nuclear magnetic resonance (NMR) derives from spin angular momentum of atomic nuclei in quantum mechanics, which has no direct equivalent in classical physics. Nevertheless, a classical mechanical description of NMR captures nearly all of the microscopic and macroscopic properties predicted from quantum mechanics and is much easier to grasp.

Atoms are characterized by mass, electrical charge, and a magnetic property called spin. Atomic nuclei that contain an odd number of protons or neutrons possess a magnetic moment, which describes the strength and direction of a microscopic magnetic field surrounding the nucleus. In the presence of a strong, constant external magnetic field, such as that produced inside an imaging magnet, a small excess fraction of polarized nuclei, on average, align themselves with the magnetic field, producing a macroscopic, measurable magnetic moment (figure 1) [9-11].

In addition, the interaction between the magnetic moment of the nucleus and the external field causes each spinning nucleus to precess (ie, change the orientation of the rotation axis of the spinning nucleus). Each nucleus precesses at a characteristic (resonant) frequency that is proportional to the strength of the external field. The resonant frequency can be calculated with the Larmor equation:

Resonant frequency F = B₀ x Larmor constant

where B₀ is the magnetic field strength and the Larmor constant (also called gyromagnetic ratio or gamma) is the specific precession frequency of the nucleus at a specified magnetic field strength. For example, a hydrogen nucleus (a proton) precesses at a frequency of 42.57 MHz per Tesla (1 Tesla is approximately 25,000 times the earth's magnetic field strength). At 1.5 Tesla, the Larmor frequency of hydrogen is 63 MHz; at 3 Tesla, 126 MHz; and at 7 Tesla, 298 MHz [12]. Because of the high abundance of hydrogen (¹H) nuclei in tissue compared with other atomic nuclei, the proton signal from hydrogen is used in virtually all clinical MRI. It is estimated that one volume element (voxel) contains 10¹⁸ water molecules.

The fraction of protons aligned with the magnetic field can be perturbed by application of radiofrequency (RF) energy at the Larmor (resonant) frequency. In essence, a transiently applied RF pulse at the Larmor frequency induces a change in the vector direction of the net magnetic moment. The density of protons as well as the rate at which the magnetization returns to equilibrium (a phenomenon referred to as "relaxation") can then be measured by the RF signal or echo emitted by the protons (referred to as "spins") as they relax. Two forms of relaxation can be measured:

●Relaxation produced by realignment of the net magnetization vector with the external field (spin-lattice or longitudinal relaxation) (figure 2A-B).

●Relaxation produced by loss of phase coherence of spins in a plane perpendicular to the external field (spin-spin or transverse relaxation) (figure 3A-B).

Phase coherence refers to a collection of neighboring spins whose magnetization vectors point in the same direction. Loss of phase coherence occurs when these vectors get out of alignment due to precession of neighboring spins at slightly different velocities, reflecting small, random variations in the local magnetic field. Both longitudinal and transverse relaxation occur at single exponential rates, with time constants T1 for longitudinal and T2 for transverse relaxation. T1 is the time constant of the exponential describing the rate of realignment with the longitudinal axis of the main magnetic field. T2 is the time constant of the exponential describing the decay of the transverse magnetization. These time constants, T1 and T2, in turn depend on the local chemical microenvironment, which varies between tissues. Therefore, three independent properties of tissue (proton density, T1, and T2) can be determined by MRI. Various methods are used to emphasize one property over another in a given MRI paradigm ("pulse sequence"), allowing tremendous flexibility in determining contrast between different tissues based on the particular pulse sequence utilized.

MRI TECHNOLOGY AND PULSE SEQUENCES — The appearance and signal intensity of tissues in an MRI depend heavily upon the type of imaging technique used.

Magnet and coil design — MRI requires a large, constant, spatially homogeneous magnetic field. Most clinical MRI systems use a superconducting magnet to generate the required magnetic field. The superconducting magnet is typically composed of windings of special metal alloys (niobium-titanium fibers in an aluminum or copper matrix) cooled by liquid helium, in which electrical current flows without loss of energy. The circular flow of electrical current in the windings induces a constant magnetic field oriented perpendicular to the windings. Additional complex windings (shim coils) are incorporated to allow application of small correction fields, improving field homogeneity. Alternative magnet technologies in clinical use include permanent and resistive magnets, which are of lower field strength than superconducting magnets.

The majority of clinical MRI systems are based on cylindrical bore magnets, in which the main magnetic field is oriented parallel to the bore and the patient placed inside the bore.

Imaging speed and resolution is defined by magnet strength, which is between 1 and 7 Tesla in most clinical scanners [12]. MRI signal is linearly proportional to the applied static B₀ magnetic field strength, while system noise is relatively independent of field strength [13]. Open or large bore scanners are available for large and claustrophobic patients and for image-guided interventional procedures. However, magnet strength for these systems are ≤1.5 Tesla, which limits image quality, although a lower field strength has the potential to decrease magnetic susceptibility artifacts when imaging the lungs [13-15]. While a higher field strength is generally desirable, an increased field strength has several problems, including increased cost and complexity of scanners, greater siting and safety issues, increased nonlinearity of radiofrequency (RF) fields, increased severity of various imaging artifacts, and higher energy deposition in tissue (heating) [16]. At higher field strengths, patients may experience vertigo, dizziness, false feelings of motion, nausea, nystagmus, and electrogustatory effects [17].

The major alternative geometry for MRI systems is the dipole magnet, in which two separate magnets of equal strength are held apart by strong metal columns, creating a gap in which the field strength is uniform. This geometry is advantageous for large or claustrophobic patients, for some interventional procedures, and for imaging off-center anatomy (eg, shoulder, elbow, or wrist), since there is much more room to position the region of interest at the center of the magnet where the uniformity is highest. The tradeoff is typically cost and upper limit of field strength, since this geometry is generally more expensive for a given field strength. Available dipole systems have maximum field strength of 1.2 Tesla in a vertical field configuration. Some of the advantages of dipole magnets, in particular for larger or claustrophobic patients, have been addressed by manufacture of cylindrical bore magnets with larger and/or shorter bores.

Four or more additional windings (or coils) are incorporated in an imaging magnet. Three gradient coils are used to create small linear perturbations of the magnetic field in any spatial directions. Because the proton precession frequency is linearly related to magnetic field strength, these magnetic gradients can encode the spatial location of protons within the body by the frequency of precession. An RF coil is used to provide the bursts of RF energy required to perturb the magnetization of spins within the body. Most MRI systems have a "whole body" RF coil permanently installed next to the gradient coils. This provides homogeneous RF excitation of a large volume of tissue within the body. Although the same RF coil can be used to receive the echo signal(s) from the body, in most cases smaller coils, tailored to different body parts, are connected via electrical cable to the RF system and placed in close proximity to the body part being imaged.

Although a full discussion of the physics of coil design is beyond the scope of this chapter, the general principles guiding coil design are:

●Closer proximity to the body part being imaged increases signal.

●Unwanted RF noise is proportional to the volume of sensitivity of a coil.

●It is possible to create one large coil out of many small coils with the low RF noise of a small coil, provided that each coil element has its own independent RF channel. Modern MRI scanners typically use multielement coils coupled to multichannel RF systems (multi-array coils), improving the signal-to-noise ratio over single channel designs. Some coils are also able to provide the RF excitation pulses, which can be helpful in limiting the overall power deposition into (heating of) the body.

The field of view prescribed in image acquisition is important. If the field of view is too small and leads to undersampling of data, aliasing and so-called wrap-around artifacts may ensue [18]. Too large a field of view decreases the spatial resolution achievable for a given length of imaging time.

The remainder of the MRI system consists of a patient transport system (the table); RF and magnetic shielding; magnetic resonance-compatible electrocardiographic and respiratory monitoring systems; RF and gradient coil power amplifiers; and a computer to set up the pulse sequences, perform collection of the echo data, and transform the raw data into images (figure 4).

Pulse sequences — Unlike computed tomography (CT), where tissue contrast depends almost entirely upon electron density, contrast in MRI is a complex function of proton density, T1 relaxation, T2 relaxation, and local chemical environment. The relative contributions of these parameters to the image vary with the manner in which the tissue is excited. The temporal pattern and shape of RF and gradient coil waveforms used to obtain an image is referred to as a pulse sequence [19]. For thoracic imaging, the most commonly used pulse sequence is called "spin echo" or "black blood imaging." For specific applications, such as imaging blood flow or rapid imaging (eg, cardiac imaging), other types of pulse sequences are employed, as described below.

Spin echo — A spin echo (SE) pulse sequence, also called "black blood imaging," typically consists of a 90-degree RF pulse to excite the tissue followed by a 180-degree refocusing pulse. The refocusing pulse reverses any dephasing effects that have occurred due to heterogeneities in the external magnetic field, forming an RF "echo" when the spins come back into phase. The elapsed time from the center of the 90-degree pulse to the peak of the echo is called the echo time (TE). The elapsed time between successive 90-degree pulses is called the repetition time (TR). A spin echo sequence with a short TR (eg, 300 to 1000 msec) and a short TE (less than 20 msec) emphasizes the T1 differences between tissues (T1-weighting). A spin echo sequence with a long TR (3000 to 6000 msec) and a long TE (greater than 80 msec) emphasizes T2 differences (T2-weighting). A long TR sequence with a short TE minimizes the effects of T1 and T2 relaxation and thus reflects proton density, which in turn primarily reflects water content. In order to increase the image contrast, inversion recovery pulses can be added.

Gradient echo — A gradient echo (GE) pulse sequence, also called "bright blood imaging," typically consists of small-angle (20 to 60 degrees) RF pulses applied in rapid succession (TR less than 100 msec). GE imaging uses a reversal of the magnetic field gradients rather than a 180-degree RF pulse to refocus spins. Image contrast in GE imaging is a complex function of TR, TE, and RF flip angle, but the imaging parameters can be adjusted to produce T1, T2, and proton density weighting similar to spin echo sequences. GE sequences generally can suffer from susceptibility artifacts related to magnetic field inhomogeneity, and disturbances in resonant frequency with subsequent loss of signal, when compared with spin echo sequences, but they are more efficient, ie, more information is acquired per unit time. Therefore, GE sequences are often used for very fast imaging or 3D volumetric imaging, in which data from an entire volume are acquired with nearly isotropic voxels, rather than the typical 2D sequences, in which the slice thickness is much greater than the in-slice pixel dimensions. Examples of GE sequences include spoiled GE imaging and balanced steady state free precession. GE imaging in the thorax is preferentially used in fast imaging, angiographic imaging, and cardiac imaging.

●With fast, spoiled-gradient imaging or single-shot imaging, images are obtained much more rapidly than with an equivalent spin echo sequence. As a result, individual images can be acquired during a single breath hold, which improves image quality due to elimination of respiratory motion artifacts. The transverse magnetization returns to zero prior to each RF pulse. In addition, the temporal pattern of enhancement of tissues following intravenous contrast material injection can be assessed by repetitive rapid imaging at one or more locations, a procedure useful in the imaging of tumors.

●Magnetic resonance angiography (MRA) uses ultrafast acquisition of T1-weighted three-dimensional GE data sets, typically obtained during a breath hold [20]. In this setting, magnetization of stationary tissue within the imaged slice becomes partially "saturated," leading to decreased signal. Blood flowing into the imaged slice, however, will not have experienced repetitive excitations and therefore will generate higher signal intensity. The pulse sequence design for contrast-enhanced MRA is based on a three-dimensional Fourier transform GE sequence, using rapid RF pulsing [20-22], eg, time-resolved imaging of contrast kinetics. The excellent contrast between the bright blood and darker stationary tissue makes GE a useful angiographic sequence for imaging of arterial and venous structures [23-25].

Balanced steady-state free precession is preferentially used for cine MRI of the heart and coronary arteries. In contrast with fast gradient imaging, the transverse magnetization reaches a steady state. Its signal is dependent on the relaxation time ratio between T2 and T1, and it provides excellent contrast between the blood and the endocardium while being independent of contrast from blood flow and its artifacts. (See "Clinical utility of cardiovascular magnetic resonance imaging", section on 'Techniques'.)

Other sequences — In an effort to achieve the most rapid imaging possible, various pulse sequences have been developed, all of which have in common the collection of a large amount of data in the form of multiple echoes following each RF excitation.

Fast spin echo is a common modification of the spin echo sequence in which multiple 180 degree refocusing pulses follow a single 90-degree excitation. Fast spin echo can reduce the time required to produce a set of images by a factor equal to the number of refocusing pulses (the echo train length) applied. Fast spin echo is often used in the thorax to allow spin echo imaging with breath holding. In the most extreme form of rapid imaging, called echo planar imaging, all of the data needed to reconstruct a single two-dimensional slice are acquired after a single RF pulse, in a time as short as 30 to 40 msec [26]. Echo-planar technique can be used for perfusion imaging of tissues after injection of gadolinium-based contrast agents (eg, in myocardial perfusion studies). (See "Clinical utility of cardiovascular magnetic resonance imaging", section on 'Pharmacologic stress CMR'.)

Ultrashort echo time (5 microsec) and zero echo time gradient echo and spin echo sequences were developed for lung parenchymal imaging in order to diminish magnetic susceptibility in the region of the alveolar walls [27,28]. (See "Magnetic resonance imaging of the thorax", section on 'Lung parenchyma'.)

Chemical shift imaging (Dixon technique) is a powerful tool for investigating the cellular composition of tissue. It takes advantage of the slight differences in resonant frequency between water protons and fat protons by using different echo times when the fat and water signals are in phase or out of phase (opposed phase). It allows for determination of the relative amount of fat and water within an individual voxel. Loss of signal intensity between in-phase and opposed-phase MRI implies intravoxel, microscopic fat or lipid. An example of the utility of chemical shift imaging is in establishing the presence of intracellular lipid in adrenal masses, which is characteristic of a benign adenoma [29]. With LAVA (liver acceleration volume acquisition)-flex sequences, rapid acquisition chemical shift imaging is obtained in a single breath hold, generating "water-only" and "fat-only" tissue imaging without increasing scan time [30].

Detection of tissue iron can be facilitated with a T1-weighted iron sequence owing to loss of signal on in-phase imaging due to excessive iron deposition and T2* effect [31]. (See "Approach to the patient with suspected iron overload", section on 'Noninvasive imaging (MRI)'.)

MAGNETIC RESONANCE CONTRAST AGENTS — Paramagnetic contrast agents, usually administered intravenously, have a role in MRI similar to that of iodinated contrast agents in CT. Most magnetic resonance (MR) contrast agents in clinical use are chelates of gadolinium. After intravenous injection, these agents initially partition into the vascular space before diffusing into the interstitium. Gadolinium, a paramagnetic element with seven unpaired electrons, influences the hydrogen atoms of surrounding tissues, resulting in an increased relaxivity with shortening of both T1 and T2; T1-shortening effect is of primary relevance at typical tissue concentrations. It is important to understand that the amplification of the signal due to increased relaxivity is not the direct effect of gadolinium but results from its indirect effect on the hydrogen atoms. On T1-weighted images, relatively greater delivery of contrast material to a region in certain conditions (eg, due to increased vascularity, slow washout of gadolinium in myocardial infarcts, increased tissue permeability in inflammation) produces an increase in T1-weighted signal intensity compared with normal tissue.

Hyperpolarized gases such as ³He and ¹²⁹Xenon can be used for detection and quantification of ventilation abnormalities and are preferentially used by some centers for monitoring of patients with cystic fibrosis [27,32]. ³He and ¹²⁹Xenon have a large magnetic moment but low natural abundance with weak signal. Through optical pumping and spin exchange, the nuclear spin polarization and MR signal are amplified, and a majority of spins align with the magnetic field [33]. Hyperpolarization of these gases improves the sensitivity and spatial resolution of MRI. The role of MRI in evaluating pulmonary disease is reviewed in detail separately (see "Magnetic resonance imaging of the thorax", section on 'Lung parenchyma'). MR contrast agent safety is discussed below. (See 'Gadolinium contrast agent' below.)

MOTION COMPENSATION TECHNIQUES — The relatively long time required to collect all the data necessary to reconstruct an image renders MRI susceptible to motion artifacts, so-called ghosting artifacts. The types of motion encountered include cardiac, respiratory, flowing blood, swallowing, peristalsis, and gross patient movements. Motion artifacts tend to be most noticeable along the axis in which position is encoded by the phase of the received echo signal. Judicious choice of the phase-encoding axis is helpful in minimizing motion artifact in an area of clinical interest. Most clinical MRI systems now have rapid imaging options, which typically can acquire all data in a breath hold, albeit with somewhat reduced spatial resolution and/or signal-to-noise ratio. This can obviate the need for motion compensation and has become common for MRI of the chest, abdomen, and pelvis. PROPELLER (periodically rotated overlapping parallel lines with enhanced reconstruction) technique is useful in reducing the number of sequences with diagnostic impairment due to motion artifact [34].

More sophisticated techniques are also available to compensate for predictable, repetitive forms of motion such as cardiac motion, respiratory motion, and blood flow.

Cardiac gating — Artifacts created by cardiac motion can be minimized if pulses are synchronized with the cardiac cycle. This requires an electrocardiographic- or pulse-monitoring system. To achieve T1-weighting, repetition time (TR) is usually set to one R-R interval. For T2-weighting (long TR), TR is chosen to be a multiple of an R-R interval. Because of the dominant effects of cardiac motion in the chest, virtually all thoracic imaging is performed with either prospective electrocardiogram (ECG) triggering or retrospective ECG cardiac gating [35].

Respiratory compensation — Normally, the phase-encoding gradient is incremented sequentially on successive radiofrequency (RF) excitations until sufficient data are collected to perform Fourier reconstruction. Respiratory compensation consists of monitoring the respiratory cycle with a flexible belt placed around the patient and reordering the phase-encoding gradients such that adjacent phase encodes occur during similar degrees of inspiration. In the reconstruction process, the data are reordered to make it appear that only one very long respiratory cycle was sampled. This greatly reduces respiratory motion artifact without changing imaging time. This type of motion compensation is therefore used routinely in thoracic imaging.

Periodic respiratory motion can be synchronized by adding a navigator pulse to the pulse sequence. This approach entails the application of a small, one-dimensional spatial-encoding gradient in a plane perpendicular to the diaphragm. It ensures that only imaging data acquired in expiration with the diaphragm in its most cranial position are utilized for the reconstruction of the image.

Flow compensation (gradient moment nulling) — Motion occurring between the application of RF excitation and reception of the echo causes changes in the phase of the received signal from the moving tissue, thereby causing misregistration of position along the phase-encoding axis. This artifact can be minimized by altering the gradient waveforms in the slice-selective and frequency-encoding axes to produce no net phase shift for constant velocity motion, a technique called flow compensation (referring to blood flow) or gradient moment nulling (GMN).

GMN produces an increase in signal within blood vessels and is therefore used routinely in angiographic (gradient echo [GE]) sequences. The GMN gradient waveforms increase the minimum achievable echo time (TE), which reduces signal-to-noise ratio in short TE (T1- and proton density-weighted images). It is generally less effective than respiratory compensation in reducing breathing motion artifact, and its use in spin echo imaging is therefore limited.

SPATIAL AND CHEMICAL PRESATURATION — Spatial presaturation refers to spatially selective prepulses applied immediately before the imaging pulse sequence to reduce signal arising from undesired spatial locations. Its main use is in reducing artifact from flowing blood on spin-echo images, but the concept is also used for suppression of respiratory motion artifact. Chemical presaturation uses a spectrally selective prepulse to reduce or eliminate signal from fat or water, taking advantage of the fact that protons bound to fat molecules resonate at a slightly different frequency than water protons, a phenomenon known as chemical shift.

Suppression of signal from flowing blood — In conventional multislice spin echo imaging, the region outside the imaged volume does not absorb radiofrequency (RF) energy, and spins from this region are therefore fully relaxed (maximum magnetic moment). If these spins flow into the imaged volume during data acquisition, ie, arterial or venous inflow, they will generate relatively high signal intensity and will also produce phase shift artifacts. (See 'Flow compensation (gradient moment nulling)' above.)

This undesired blood flow artifact can be suppressed by presaturating the nonimaged volume with a 90-degree RF excitation and randomizing the phase of these spins with a large magnetic field gradient, at a small cost in increased imaging time. Spatial presaturation is routinely used in spin echo thoracic imaging sequences to reduce blood flow artifact.

Suppression of respiratory motion artifact — Suppression of respiratory motion artifact can be achieved by presaturating the anterior chest wall to reduce its emitted signal. This type of saturation within the imaged slice is used mainly in conjunction with surface phased-array coils. These receiver coils greatly increase the signal-to-noise ratio obtained per unit of imaging time; however, they suffer from nonuniform spatial sensitivity. Thus, signal from protons close to the coil, ie, the chest wall, is received with greater efficiency than signal from protons deep within the thorax. Spatial presaturation can deemphasize the signal from the chest wall and therefore reduce respiratory artifact arising from the chest wall as well. The main limitation of spatial presaturation for this application is that only a linear band of tissue can be suppressed with a single pulse. To suppress signal from the entire chest cage, multiple presaturation pulses must be applied, at a greater cost in imaging time.

Navigator gating of the diaphragmatic movements can effectively freeze respiratory motion while the patient is freely breathing and is a newer technique for suppression of respiratory motion artifacts.

Single-shot sequences acquire each section during a single breath hold, with data acquired during each heartbeat or every other heartbeat in cardiac imaging.

Suppression of signal from fat — As noted above, chemical shift fat suppression is commonly used to reduce the high signal intensity of fat-containing tissues. This comes at a modest cost in imaging time. Chemical fat suppression (CHEM SAT) is effective but relies on a high degree of magnetic field homogeneity within the imaged volume: Since the difference in resonant frequency of water and fat protons is only 3.5 parts per million (at 1.5 Tesla), any variation in the magnetic field of this order or larger will cause uneven fat suppression. The problem is exacerbated on lower field strength imaging systems since the fat-selective prepulses have a finite bandwidth that is more likely to overlap the water resonance frequency due to the very narrow difference in resonant frequencies.

A common alternative method of fat suppression is known as short tau inversion recovery (STIR), which exploits the fact that the relaxation time of fat protons is much shorter than water protons. By exciting all tissue within a volume to invert the magnetization (180-degree inversion or presaturation pulse), followed by a delay chosen when fat protons have no net magnetic moment, followed by a 90-degree pulse and then readout pulses, signal from fat can be suppressed without requiring high field homogeneity. The main disadvantage of this method over CHEM SAT is that it is slower.

Use of fat suppression within the thorax enables tissue characterization when a fat-containing lesion is suspected (either method effective) and facilitates differentiation of fat from contrast enhancement after intravenous contrast administration, where enhancing tissue and nonsuppressed fat may approach equal signal intensity (only CHEM SAT effective).

Flow-Independent Dark-blood DeLayed Enhancement (FIDDLE) technique allows visualization of tissue contrast enhancement while suppressing blood-pool signal and is useful in cardiac imaging with late gadolinium enhancement [36].

PRACTICAL ASPECTS — During the MRI examination, the patient typically lies supine on a sliding table, which is advanced into the bore of the magnet for imaging. Since noise levels can be high during imaging due to vibration of the gradient coils, earplugs or specially designed in-ear audio systems are commonly used. An imaging sequence may be as short as one second or as long as 10 minutes, and the entire examination typically lasts 20 to 60 minutes as multiple imaging sequences are performed in multiple planes. More limited and rapid MRI examinations are sometimes used in the setting of trauma or with a patient unable to tolerate a more complete examination for whatever reason.

PRECAUTIONS — Conventional clinical MRI is rarely associated with any adverse effects. Most contraindications are relative precautions, which can be divided into five groups: implanted devices and foreign bodies, unstable patients, pregnancy, gadolinium contrast agents, and other.

Implanted devices and foreign bodies — Prior to scanning, a thorough safety screen of each patient for any implanted, embedded, or attached devices or foreign objects is mandatory. These considerations and associated screening procedures are discussed elsewhere. (See "Patient evaluation for metallic or electrical implants, devices, or foreign bodies before magnetic resonance imaging".)

Unstable patient — Because of difficult access to a patient within the magnet and because most resuscitation equipment cannot be brought into the scanning room, an unstable patient probably should only undergo MRI for an urgent clinical indication without another acceptable imaging alternative.

Pregnancy — The risks posed to the developing fetus by MRI are unknown. These risks, if any, are probably low and use of MRI may help avoid ionizing radiation associated with other imaging modalities. Gadolinium-based contrast agents cross the placenta and are not recommended for use in pregnant patients unless the potential benefit justifies the potential risk to the fetus. MRI is indicated for use in pregnant women if the potential benefit outweighs the potential risk. (See "Diagnostic imaging in pregnant and lactating patients".)

Gadolinium contrast agent — The major route of MRI contrast excretion is renal. However, unlike iodinated contrast media, MRI contrast agents rarely cause anaphylactoid reactions. (See "Diagnosis and treatment of an acute reaction to a radiologic contrast agent", section on 'Gadolinium-based contrast'.)

Patient safety considerations in giving gadolinium-based contrast agents for MRI are discussed elsewhere. (See "Patient evaluation before gadolinium contrast administration for magnetic resonance imaging" and "Diagnosis and treatment of an acute reaction to a radiologic contrast agent".)

Other — Several other problems may occur:

●Some patients may experience claustrophobia when asked to lie within an MRI scanner. Although estimates vary and the prevalence is likely a function of culture, in the largest study including over 55,000 patients, the rate of MRI-induced claustrophobia necessitating intervention was 2.1 percent [37].

For some patients, claustrophobia may be severe enough to require sedation. Various remedies, including magnetic resonance-compatible music and video systems to "distract" the claustrophobic patient, and acoustic insulation to reduce the loud noise from the energizing/de-energizing of radiofrequency and gradient coils, are commercially available. In addition, a large number of so-called "open" MRI systems have been installed to cater to patients who experience claustrophobia or are too large for a cylindrical bore magnet. Furthermore, small cylindrical bore MRI systems are available for imaging of the upper or lower extremities (excepting the shoulders and hips), which allow the rest of the body to remain outside of the magnet.

●An agitated patient or one who cannot hold still within the magnet is a poor candidate for MRI unless sufficiently sedated; motion artifacts resulting from gross patient movement are typically severe enough to interfere with accurate diagnostic imaging.

●Some obese or large patients may not be able to fit into the bore of a typical cylindrical magnet. Such patients can usually be accommodated by an open MRI system or by some of the wide-bore MRI scanners.

SUMMARY

●MRI physics and technology – MRI is an imaging technology that uses nonionizing radiofrequency (RF) radiation inside a strong magnetic field to detect the location and local chemical environment of protons. The imaging process can be summarized as induction of protons by a magnetic field, followed by excitation with RF pulses, subsequent relaxation, and eventual readout with RF receiver coils. (See 'Magnetic resonance physics' above and 'MRI technology and pulse sequences' above.)

●Enhanced visualization of tissue contrast – Unlike CT, which detects only differences in electron density, the tissue contrast that can be achieved with MRI is extremely flexible and orders of magnitude greater. (See 'Magnetic resonance physics' above.)

●MRI contrast agents – As with CT, intravenous injection of MRI contrast agents is often used to improve contrast between pathologic and normal tissues or to perform angiography. (See 'Magnetic resonance contrast agents' above.)

●Appropriate patient selection – Limitations of the use of MRI include hazards posed by certain indwelling metallic devices (for which safety references are available) and claustrophobia. So-called open MRI scanners and short-bore conventional scanners can accommodate most claustrophobic patients, some of whom may still require sedation to complete the imaging study. (See 'Practical aspects' above and 'Precautions' above.)

●Safety – Conventional clinical MRI is rarely associated with any adverse effects. Most contraindications are relative precautions, which can be divided into five groups: implanted devices and foreign bodies, unstable patients, pregnancy, gadolinium contrast agents, and other. (See 'Precautions' above.)