Dr. Draud is currently Chief Resident in Diagnostic Radiology at Vanderbilt University Medical Center, Nashville, TN. She earned a double major in Biomedical Engineering and Mathematics, graduating cum laude, from Vanderbilt University in 1994. She subsequently received her MD from the University of Kentucky, Lexington, KY, and graduated with distinction. Dr. Draud will begin a fellowship in musculoskeletal imaging at the University of California, San Diego, in 2003.

Magnetic resonance (MR) imaging has advanced at an astonishing speed since its inception in the early 1970s. Although the overwhelming majority of examinations are safe and uneventful, injuries and rare fatalities continue to occur. It is important to recognize and understand that there are three distinct fields inherent in MR imaging to be evaluated for bioeffects and potential safety issues: the static magnetic field, the radiofrequency electromagnetic field, and the time-varying gradient magnetic field. The static magnetic field has no known upper limit for safe exposure, although increasing use of high-field imaging spotlights the need for continued research. The radiofrequency electromagnetic field is limited by induced currents in electrically conductive material and can lead to burns. The time-varying gradient magnetic field can cause nerve and muscle tissue stimulation by induction of electric fields. Effective screening includes both written questionnaires and verbal interviews. Intensive screening of biomedical implants and devices as well as for foreign metallic objects is a critical part of MR imaging safety.

Although the safety and bioeffects of magnetic resonance (MR) imaging have long been of interest to those in radiology, recent events and research have brought this topic to the forefront of medical imaging. MR safety and bioeffects are currently receiving significant attention from the research community as new technologies are being developed and brought to the patient care arena. Equally as important is the spotlight on patient safety from the general medical community and the public, which manifested most significantly after a 6-year-old boy received a fatal head injury from a projectile metal oxygen tank in an MR imaging suite in July 2001. ¹ This tragic occurrence was reported both by The New York Times and The New England Journal of Medicine and garnered international attention. ^1,2

Since the yearly estimate for MR imaging is more than 8 million studies in the United States and more than 20 million worldwide, it is clear that the procedure is uneventful and safe in the overwhelming majority of cases. ^1,3 However, there are hazards associated with MR imaging that can threaten safety. In addition to the previously mentioned death, a fatality resulted from the torquing of an aneurysm clip, and implantable medical devices, such as cardiac pacemakers, have been implicated in 5 more deaths. ^3,4 Other injuries include unilateral blindness because of magnetic force on an intraocular metallic foreign body and burns from external monitoring devices, such as pulse oximeters. ^5,6 The numerous examples of objects being drawn into the magnet bore are perhaps most disturbing and can also be quite hazardous, even fatal, as in the instance of the oxygen tank.

These occurrences--along with frontier research involving high-field imaging, rapidly switching magnetic fields, radiofrequency (RF) electromagnetic fields, and new advances in MR imaging intervention--contribute to the importance of the topic of MR imaging safety and bioeffects. Efforts such as the recent American College of Radiology White Paper on MR Safety, ⁷ an educational course at the International Society for Magnetic Resonance in Medicine annual meeting, a categorical course at the 2001 annual Radiological Society of North America meeting, and development of Web sites (eg, MRIsafety.com), are attempts to highlight the significance of the topic. ⁷ This review is intended to serve as an aid in practical day-to-day patient care for physicians at different points in their training and careers.

MR bioeffects

There are three distinct fields inherent in MR imaging that have been evaluated for bioeffects and potential safety issues (Table 1). The static magnetic field, B ₀ , is currently limited only by technology and cost because no upper limit for safe exposure to the intense static field has been shown. ³ The RF electromagnetic field, B ₁ , is limited by thermogenic qualities of RF radiation. ⁸ Finally, the time-varying gradient magnetic field can affect peripheral nerve stimulation (PNS). ⁹ These three fields along with acoustic considerations are regulated by the U.S. Food and Drug Administration (FDA) Center for Devices and Radiological Health guidelines, which are summarized in Table 2. ¹⁰ The International Electrotechnical Commission describes a three-tiered safety system of increasing restriction that consists of a normal operating mode, first controlled operating mode, and second controlled operating mode, as described in Table 3. ^11,12

Bioeffects of static magnetic fields

The static magnetic field produces a net magnetization of the protons in a patient's tissue, which causes the protons to be aligned with the field. The strength of the field may vary and is determined by the specific type of magnet used. In a comprehensive review article on the safety of strong, static magnetic fields, Schenck ³ summarizes the investigative work performed to date as lacking "a single example of a scientifically sound and rigorously verified pathologic effect of such [static] fields." He attributes this lack of detrimental effects to the absence of in vivo ferromagnetic components in healthy subjects and to the extremely weak diamagnetic susceptibility of human tissue. It is important to note, however, that technology is advancing rapidly and safety concerns need to be revisited at higher static magnetic fields. High static magnetic fields are loosely defined as >=1.0 T. ¹³ Imaging of humans has already been performed at 8 T for research purposes. ^14,15 A 9.4 T magnet is now being built for human studies, and field strengths of 12 T and higher may be possible with the development of flux pumped magnets. ¹⁶

Well-known nonpathologic biological effects of the MR imaging static magnetic field include elevation of the T wave in electrocardiographic (ECG) tracings. ¹⁷ The movement of blood, a conductive medium, through the static magnetic field, B ₀ , causes a magnetohydrodynamic effect that produces a voltage across the vessel. Because the maximal flow rate of blood occurs during ventricular contraction, or the T wave interval, this electromotive force (EMF), or added voltage, is seen as an elevation of the T wave (Figure 1). This effect was an important consideration in the original evaluation of MR safety as well as in the initial work with higher static magnetic fields. However, no detrimental outcomes from the expected T wave alterations were observed at exposure to a static magnetic field of 8 T in human or porcine subjects. ¹⁴ This induced EMF, although shown to be safe at 8 T, should continue to be evaluated for potential biologic effects.

Other reversible, nonpathologic effects occurring at high field imaging include vertigo, metallic taste, and magnetophosphenes or visual flashes. ¹⁸ These effects can be reduced with slow movement across magnetic field gradients to minimize induced electromotive forces.

Bioeffects of RF electromagnetic fields

The main bioeffect of exposure to RF radiation is related to the thermogenic qualities of the electromagnetic field. ⁸ Deposition of energy in the body because of RF radiation is quantified by determining the specific absorption rate (SAR). The SAR is the mass normalized rate at which RF power is coupled to biological tissue and is expressed in units of watts per kilogram of body weight (W/kg). The FDA guidelines for SAR limits are shown in Table 2. ¹⁰

Because of resistive losses, most of the transmitted RF power is transformed into heat within the patient's tissue. ⁶ Studies have been performed up to a whole-body SAR of 6 W/kg and showing no deleterious effects in an individual with normal thermoregulatory function. ⁶ Recent research has suggested that there may be an in vitro RF-induced cell membrane target effect with an increase in cytosolic calcium concentration. ¹⁹ The significance of this is unclear and further work is necessary to support or refute this data, as well as apply it clinically. The use of more "RF-intensive" MR imaging protocols using fast-spin-echo (FSE) and magnetization transfer contrast pulse sequences, especially with high-field strength systems, will require continuing investigation into their safety. Also, patients who have an underlying health condition or are taking medication that might impair their thermoregulatory system will require further evaluation.

In addition to direct RF heating of the body, more than 150 incidents have been reported involving burn injuries to patients undergoing MR imaging examination resulting from inductive heating of electrically conductive devices and materials placed in contact with the body. ²⁰ First-, second-, and third-degree burns have been associated with ECG leads, pulse oximetry, and imaging coils. ²⁰ Less frequently, ferromagnetically active pigment in permanent eyeliner, a breast implant metal-backed expander injection portal, and other devices have been reported to show heating. ²⁰

Heating mechanisms include direct electromagnetic induction, resonant circuit, and antenna effect. ²⁰ Direct electromagnetic induction can be minimized by preventing the looping of conductive cables and placing electrodes as close together as possible. Other precautions include placing the lead and cable away from the RF coil and using fiberoptic or high-resistance graphite electrodes and cables. ²¹ An inadvertent conductive loop can be formed on a patient between two points by touching the arms or legs. All skin-to-skin contact points should be eliminated, and cables should be insulated from the patient's skin (Figure 2). An additional degree of safety can be achieved by using foam padding to insulate the patient from all electrical components. Maximum current induction will occur when a circuit is in a resonant condition. ²¹ The "resonant frequency" occurs when the tissue is approximately half the size of the incident wavelength. ⁶ Burns can also occur with straight wires, as with pulse oximetry leads from the antenna effect. The antenna effect results when lengths of conductive cable are used and an additional electric field is induced with a maximum point at the tip. ²²

Bioeffects of time-varying gradient magnetic fields

Pulsed gradient magnetic fields are used in MR imaging to produce resonant frequencies that encode the data spatially. These time-varying magnetic fields, dB/dt, can induce electric fields in patients, ²³ which, in turn, can lead to nerve and muscle stimulation.

The mean threshold levels are 60 T/s for the peripheral nerve sensation;
90 T/s for peripheral nerve pain; 900 T/s for the respiratory system; and 3600 T/s for the heart. ¹⁷ Currently, the FDA does not mandate a numerical upper limit value for dB/dt. Rather, dB/dt is limited by "severe discomfort" or painful nerve stimulation. ¹⁰ When gradient ramp times are less than 1 to 2 millisecond, the probability of cardiac stimulation is estimated to be near zero, providing dB/dt levels are low enough that PNS is rare. ²³ However, the widespread use of echoplanar imaging (EPI), allowing faster acquisition of images with rapid switching of magnetic gradient fields at higher amplitude, has increased the incidence of PNS. ²⁴ Recently, another frequently used fast sequence, steady-state free precession, was found to have threshold parameters similar to EPI. ²⁵

Peripheral nerve stimulation can be uncomfortable for patients, limiting their cooperation or the quality of MR imaging. With continued advances in rapidly switching gradient magnetic fields, this muscle and nerve stimulatory effect will remain an area of concern.

Acoustic noise

Various types of acoustic noise are produced in an MR examination. The frequently described tapping or knocking noises of the MR scanner are mainly due to the pulsed gradients. The rapid switching of currents within the large static magnetic field produces strong Lorentz forces that physically deform the gradient coils. ²⁶ These coils impact against their mountings, which also flex and vibrate, causing acoustic noise that is enhanced with decreased field of view, smaller section thickness, and shorter repetition times. ²⁷ Echoplanar imaging and FSE sequences use high gradient amplitudes and very fast gradient switching times that, in turn, generate higher levels of acoustic noise. ²⁷

The FDA mandates that acoustic noise levels of MR systems not exceed a peak acoustic noise of 140 dB. ¹⁰ Even before EPI and FSE sequences, temporary shifts in hearing thresholds were reported in 43% of patients scanned with improper or no hearing protection. ²⁸ The routine use of earplugs or headphones for patients or staff who might be in the MR system for long periods of time is recommended. ²⁷ Unfortunately, these can impede communication with the patient, so "antinoise" or active noise cancellation systems are being developed. ²⁷ Noise levels have also been reduced by a number of different sound-"damping" methods such as the installation of acoustic foam around the gradient coils. ²⁶

Pregnancy

Imaging of pregnant women has long been fraught with concern. Although MR imaging lacks ionizing radiation, safety considerations of the static, gradient, and RF electromagnetic fields warrant caution. Currently, there is insufficient data supporting or refuting the use of MR imaging during pregnancy. The MR safety committee of the Society for Magnetic Resonance Imaging indicates that MR can be performed on pregnant women in certain circumstances: when other nonionizing forms of imaging such as sonography are inadequate or when the examination provides important information that would otherwise require exposure to ionizing radiation. ²⁹ Patients should be informed that while no data have shown deleterious effects, MR safety has not been proven. ³⁰ Written consent should be obtained.

Pregnant health-care workers present a different set of concerns. Exposure to RF electromagnetic or gradient fields is very limited in this population because these fields are present basically only within the system bore. However, the exposure to the static field may be chronic and low level. An epidemiological study of imaging technologists failed to show any statistically significant elevation in the rate of spontaneous abortion, infertility, or premature delivery. ³¹ Shellock and Kanal ²⁹ recommend a policy to "permit healthcare workers to perform MR procedures, as well as to enter the MR system room, and to attend to the patient" during pregnancy, regardless of trimester. They also recommend, as a precautionary measure, that the pregnant worker exit the MR system room during operation of the system. ²⁹

MR safety

Ensuring a safe workplace environment should be the top priority for all involved with MR imaging. With this goal in mind, the task of providing and maintaining a secure setting can be a daily challenge. The first step involves having a set of safety guidelines in place as a framework in which to operate. Second, the screening of both patients and healthcare workers is extremely important. For patients, a combination of written and verbal screening is common (a sample screening form is presented in Figure 3). ³² Notably, a history of uneventful MR imaging does not preclude the screening process at the time of subsequent examinations. Differences in magnetic field strength, patient positioning, and gradient-switching fields in each examination can change the circumstances significantly.

Safety of biomedical implants and devices

Evaluation of implants and devices is critical to patient safety in the imaging milieu. The FDA defines two terms for items being considered for the MR environment. "MR safe" indicates that the device presents no additional risk to the patient in the MR surroundings but can affect the image quality. "MR compatible" indicates that the device presents no additional risk to the patient in the MR environment and does not affect image quality. ¹⁰ It is important to note the magnetic field strength of the system used during testing, particularly with the increasing use of higher field strength magnets. For example, a device shown to be "MR safe" at 1.5 T may not necessarily be safe at 3.0 T.

Metallic biomedical implants and devices with ferromagnetic properties are subject to torque and translational force. Torque is a rotational force and results in a ferromagnetic object aligning parallel to the static magnetic field. Torque is greatest in the center of the magnet, but also exists outside the magnet bore. Translational force is a linear force that attracts an object into the magnet bore. This force is greater the closer it is to the magnet bore where a higher spatial gradient exists. As an object gets closer to the magnet bore, the spatial gradient goes higher, and greater translational force is exerted on the object. Once the object is in the center of the magnet, there is no longer a spatial gradient, and thus no translational force.

For metallic implants with significant ferromagnetic properties, these forces can result in tissue damage surrounding the apparatus, including rupture of blood vessels. Examples of potentially injurious implants include cerebral aneurysm clips and intraorbital metal fragments. The fatality involving the ferromagnetic aneurysm clip occurred with the development of intracerebral hemorrhage and tearing of the middle cerebral artery while in the MR unit. ⁴ Imaging suggested that torquing of the clip was the cause. The reported case of unilateral blindness resulted from an occult 2.0 * 3.5 mm intraocular metal fragment dislodging in a 0.35 T scanner causing vitreous hemorrhage. ⁵

Radiofrequency electromagnetic fields can cause the heating of an implant, its lead wires, or surrounding tissues. ³³ An example is RF energy interacting with a pacemaker output potentially causing tachycardia or other arrhythmias.

Imaging with an implantable device requires thorough investigation regarding the MR safety or compatibility of the item with attention to the testing environment, ie, field strength. Detailed information for the specific devices appears in the Guide to MR Procedures and Metallic Objects: Update 2001 ³⁴ or at MRIsafety.com. ³²

The projectile effect

Torque and translational force are also responsible for the projectile effect or missile effect in which unrestrained magnetic objects can literally fly through the air and crash into the magnet. ¹³ Even a very small object can accelerate and cause injury to patients or others within the area, as well as to the MR system itself.

A wide variety of objects, both medical and otherwise, have been drawn into magnets. Such items include oxygen or nitrous oxide tanks, a defibrillator, a wheelchair, a respirator, ankle weights, an IV pole, a toolbox, sandbags containing metal filings, a vacuum cleaner, and mop buckets. ³⁵ Other reported objects include parts of a forklift, a laundry cart, a chair, a ladder, a light fixture, scissors, traction weights, a pillow with ferromagnetic springs, and more. ^36,37 Recently, there has even been a report of the spontaneous firing of a gun into the wall of the imaging room despite the firearm's thumb safety being engaged. ³⁸ The projectiles can do tremendous damage to patients and to the MR systems. Figure 4 presents an example of the IV pole that was inadvertently drawn into the bore of a 1.5 T magnet.

These reports of metallic objects leading to adverse events are particularly worrisome because they are largely preventable with a combination of education and vigilance. A healthcare risk management article published after the projectile oxygen tank caused the child's death suggested that at least every 2 weeks, MR imaging manufacturers across the country remove metal objects from magnets. ³⁶ The Emergency Care Research Institute has issued several recommendations to help prevent these incidents, including appointing a safety officer to enforce procedures, providing formal safety training to those working in the MR environment, and restricting the equipment within the 5-gauss line to that labeled "MR safe." ³⁹

American College of Radiology (ACR) White Paper on MR safety

The ACR Blue Ribbon Panel on MR Safety recently issued a report after the MR imaging incident and subsequent media attention spotlighted the need for review. ⁷ The committee was chaired by Emanuel Kanal, MD, FACR, a leader in MR safety. In the editorial, Rogers states that "safety in the MR suite must be a collaborative effort." ⁴⁰

Guidelines include establishing safety policies and designating a person responsible for ensuring their implementation. A plan for site access restriction using four zones is also outlined. Zone I is the general public access area. Zone II is an interface to the restricted areas for the initial screening and checking-in of patients. Zone III is restricted and is controlled by MR personnel, usually technologists, under the authority of the MR medical director. This zone should be physically restricted with access keys/passkeys needed for entry. The 5-gauss line is within this zone and all persons entering Zone III must have passed the screening process. Zone IV is the room in which the MR system is located. ⁷

The ACR article describes levels of MR personnel, which are based on the amount of safety education and training they have received. It states that when scanning in a nonemergency coverage situation, there should be at least two MR technologists (or one MR technologist and one other individual from MR personnel) in the MR environment. ⁷

In regard to pregnancy, the committee stresses that to scan a pregnant woman, the data needed should "affect the care of that patient and/or fetus DURING the pregnancy." Other guidelines address contrast agent safety, cryogen-related issues, the bioeffects of magnetic fields, and other topics. ⁷

Particular attention is paid to screening in the ACR White Paper. The guidelines suggest that nonemergent patients be screened safely by a minimum of two separate individuals, with one screening being verbal. Any non-MR personnel also must pass the screening process to enter Zone III and, by definition, Zone IV, as it is contained within Zone III. Other specific guidelines address the screening of devices and review questions of clinical concern, such as when to obtain orbital plain films. ⁷

In a commentary following the ACR White Paper, Shellock and Crues ⁴¹ described the concept of zones as interesting but without empiric support for its use. Additionally, they felt it was probably unnecessary to mandate screening by two different individuals for nonemergent patients. Other areas of disagreement included the use of a strong, handheld magnet to test devices or implants and the use of color-coded labeling designating products as "MR safe" or "MR compatible." A formal response to this commentary will be published in an upcoming issue of the American Journal of Roentgenology .

Conclusion

The last 2 decades have provided an explosion of exciting technology and applications for MR imaging. Unfortunately, there have also been injuries and fatalities associated with MR examinations. Safety in the MR environment has long merited the attention it is now receiving. The radiology community has highlighted MR safety and bioeffects with journal articles, meeting courses, Web sites, and the new ACR White Paper on MR Safety. With technological advances providing higher static magnetic fields and more rapidly switching gradient magnetic fields, research to delineate the bioeffects of MR imaging is of paramount importance. Ensuring a safe environment is part of the fundamental commitment we make to our patients. Awareness, vigilance, and resolve can help make this challenge a reality.

Acknowledgement

The author thanks Ron R. Price, PhD, for reviewing the manuscript and for providing invaluable insight. The author is also thankful for thoughtful assistance from the "MR King," Ric Andal, RT(R)(MR), and for technical support from John Bobbitt. Special thanks are extended to Michael Rohmiller, MD for his generous assistance in obtaining the images that appear in this article.