MRI Safety Tutorial

Table of Contents

Disclaimer: While every attempt has been taken to verify the information contained in this document, the authors cannot guarantee its accuracy. This document should not be construed to represent a definitive interpretation of the regulatory statutes regarding MRI, and the reader should be aware that regulations might change and render obsolete some information specified herein. This document may not be distributed or reposted without the express written permission of the BIAC director.


MRI utilizes a very strong static magnetic field, time varying smaller magnetic fields ("gradients") and an electromagnetic, radio frequency field to create images. The static field (B0) is the strong magnetic field created by the superconducting coils and is usually measured in units of Tesla (a Tesla (T) is equivalent to 10,000 gauss. In comparison, the earth's magnetic field roughly .00005 T, so a 1.5 T scanner has a static field that is approximately 30,000 times stronger than the earth's magnetic field). As its name suggests, the static field is time invariant and is always on. However the two remaining fields, the radio frequency electromagentic field (B1) used to excite the spins, and the gradient magnetic fields used to spatially encode, are rapidly turned on and off during scanning. Each of these three components pose different concerns with regard to subject safety, and each will be discussed in detail in subsequent sections.

In addition to safety issues related to the static and time varying magnetic fields and radio frequency energy, the use of coils to create these fields imposes a particular geometry upon the device itself. The long narrow magnet bore and special RF coils create subject comfort and compliance issues - particularly for subjects who experience claustrophobia in the confined space. Finally, the ramping up and down of the gradient fields (by applying varying currents to the gradient coils) causes them to vibrate due to Lorentz forces, and this movement causes the loud noises associated with scanning.

When compared to imaging modalities that employ ionizing radiation such as x-ray and positron emission tomography, magnetic resonance imaging (MRI) is a safe modality. In a recent review, Schenck (2000) estimated that 150,000,000 MRIs were performed worldwide between the inception of widespread clinical testing in the early 1980s and 1999, that 20,000,000 are performed each year, and that more than 50,000 scans are performed each day. That the vast majority of these scans are performed without incident is a comforting fact. However, the very serious exceptions to this generalization should give pause. Schenck (2000) reports that his review of the literature in 1998 showed that seven deaths occurred that were attributable to MRI scanning, most related to MRI-induced malfunctions in cardiac pacemakers. A study by Chaljub (2001) detailed five serious accidents involving ferromagnetic projectiles at two MRI facilities in Texas. A recent projectile accident caused the death in New York of a six-year old boy in a MRI accident, and has focused the public's attention upon MRI safety. Only through constant vigilance and strict adherence to standard operating procedures can serious accidents be avoided.

Static Field (Bo)

The main safety risk from the static field results from the translation and torsion experienced by ferromagnetic objects within the magnetic field. This can lead to movement or malfunction of implanted medical devices and metal debris, and the acceleration into the bore of unsecured metal objects (projectiles).

The best method to counter this risk is a comprehensive screening of all subjects and staff for ferromagnetic objects within their bodies or on their persons prior to their entry into the magnet room. Constant vigilance and testing is required to keep the MRI suite free of unsecured ferromagnetic objects that might become dangerous missiles.

Influence of static field on metal objects (projectile effect)

Objects that are constructed in part or whole with ferromagnetic materials (iron, nickel, cobalt, and the rare earths chromium, gadolinium, dysprosium) will be strongly attracted to the magnet bore. Steel objects are highly ferromagnetic, and some medical grades of stainless steel are ferromagnetic. Even non-magnetic forms of stainless steel can be made magnetic if the material is worked. Metals such as aluminum, tin, titanium, gold and lead are not ferromagnetic, but objects are rarely made of a single metal. For example, ferromagnetic steel screws may secure titanium frames for glasses. Always check unknown objects with a permanent magnet to determine if they are ferromagnetic.

A ferromagnetic object will experience a magnetic pull that increases greatly as it approaches the magnet bore. A movement of just six inches towards the bore of the magnet can nearly triple the force experienced by the object, making it impossible to hold on to a ferromagnetic object such as a wrench or screwdriver. Similarly, a pager may stay clipped to your belt as you stand at the end of the gurney, but be propelled into the magnet bore at 20-40 mph if you bend over the bed. A subject in the bore of the magnet could be seriously injured, or even killed, by such a projectile. Even if not injured, such an accident could be emotionally traumatic. Chaljub (2001) reported that the family of a patient who narrowly escaped injury when a ferrous oxygen tank smashed into a magnet sued and was awarded $35,000 for psychological duress.

The following two incidents reported to the FDA illustrates the serious consequences associated with projectiles:

It is absolutely essential to remove all ferromagnetic objects prior to entering the magnet room, including pagers, PDA, cell phones, stethoscopes, pens, watches, paperclips, and hairpins. It is absolutely essential that all such objects be removed from the subject, as he or she will experience a much greater static magnetic field in the center of the bore. Careful screening of subjects following BIAC Standard Operating Procedure (SOP) is essential.

It is also essential to test all equipment brought in, or near, the magnet room. A permanent magnet can be used for this purpose. The BIAC magnet rooms are equipped with swinging screen doors to prevent metal carts from gradually rolling into the magnet room. Never prop these doors open.

You should never leave the door to the magnet room open while a subject is being set up for scanning. A routine of keeping the door shut will prevent the entry into the magnet room of unauthorized and unscreened individuals - such as maintenance personnel who may be carrying tools - while you and the subject are in a vulnerable position. Always challenge unknown persons who enter the console room to determine their authorization for being there. Better to be embarrassed than to permit an accident to occur.

Influence of static field on implanted metal devices or metal debris


Ferromagnetic devices and debris will attempt to align parallel with the static magnetic field. This effect is dominant when the static field is spatially invariant. Thus, ferromagnetic vascular clamps such as those used to clamp an aneurysm might rotate slightly in the field and cause bleeding. The following example reported to the FDA illustrates the serious consequences.

Torsion effects have also been used to explain the swelling and/or irritation that has been reported for subjects with tattoos and certain makeup - particularly mascara and eyeliner. The pigments in tattoos and makeup may contain iron oxides in irregular shapes that attempt to align with the magnet field and produce local tissue irritation(2000)


Ferromagnetic devices and debris within a subject's body will be attracted to the magnet, just as would unsecured objects in the room. Thus, translation effects are similar to projectile effects described above. The following examples reported to the FDA illustrate problems associated with metal shards in the patient's eye moving in the magnetic field and causing serious injury.

Influence of static field upon human physiology

In a recent review, Schenck (2000) has reviewed the literature concerned with the influence of static fields upon human physiology for magnets with field strengths up to 8 Tesla. He concluded that "In the absence of ferromagnetic foreign bodies, there is no replicated scientific study showing a health hazard associated with magnetic field exposure and no evidence for hazards associated with cumulative exposure to these fields."

Schenck's (2000) optimistic conclusion pertains to long lasting health risks associated with static magnetic fields. There have been anecdotal reports of less serious and short-lived effects associated with static field strengths greater than 2 Tesla. These include reports of visual disturbances ("phosphenes"), metallic taste sensations, sensations in teeth fillings, vertigo, nausea, and headaches. These sensations occur infrequently, but appear to occur when the individual's head is moved quickly within the static field.

Magnetohydrodynamic effects

It is believed that some of these effects — particularly vertigo, nausea, and phosphenes — may be related to magnetohydrodynamic phenomena. When an electrically conductive fluid, such as blood, endolymph fluid, or aqueous fluid flows within a magnetic field, an electric current is produced, as is a force opposing the flow. For example, within blood vessels, the potential across such a vessel is

E = 0.1 B0vd, where

E = potential across the vessel in mV
B0 = magnetic flux density (in Tesla)
v = blood velocity (cm/s)
d = blood vessel diameter (cm)

This potential is negligible in all but the largest arteries, such as the aorta where values on the order of 5 mV/T can occur. This can cause apparent abnormalities in a subject's electrocardiogram, such as an enlargement of the T-wave, but this only an artifact and does not reflect evidence of an alteration in the heart's rhythm or activity.

A force is generated that either opposes or accelerates the magnetohydrodynamic force. In the case of blood flow, the magnetohydrodynamic force is resisted by an increase in blood pressure. However, this effect is negligible, requiring 18 Tesla to generate a change of 1 mm/Hg in blood pressure. These resistive forces could, however, impose torque upon the hair cells in the semicircular canals causing vertigo and nausea, or upon the rods or cones in the retina causing the sensation of phosphenes. We emphasize that these latter effects are likely only to occur during quick movements of the head within the field. Moving the subject slowly in and out of the scanner and restricting head movement should eliminate these sensations.


At most research MRI facilities, including BIAC, potential subjects who are pregnant must be excluded from participation in research MRI studies unless specific IRB approval has been obtained. As no technology can be proven safe, this policy is prudent. However, it should be emphasized that at present there are no known risks to imaging during pregnancy, either for the woman or the fetus.

Kanal and colleagues (1993) conducted an epidemiological study to assess possible deleterious effects of MRI on the reproductive health of female MRI technicians. There were no significant effects of working with MR on rates of miscarriage, stillbirth, spontaneous abortion, or ectopic pregnancy among the 1,915 women responding to the survey. In addition, no effects on rates of premature delivery, infertility, or low birth weight were found.

Studies measuring the effects of MRI upon human cell development have also failed to find evidence of abnormalities attributable to MRI. Wiskirchen and colleagues (1999) exposed fetal lung fibroblasts to a 1.5 T scanner regularly for three weeks, and found no significant differences in proliferation between imaged and control cell cultures. Rodegerdts and colleagues (2000) exposed fetal human lung fibroblasts to time-varying magnetic fields (such as those produced by gradient coils) and found no significant differences between exposed and control cells. In addition, they found that neither the strength of the field nor the amount of exposure contributed to any teratogenic effects.

MRI is being used increasingly for clinical assessment of unborn fetuses. No evidence of stress to the unborn fetus has been found during MRI scan acquisition. For example, Poutamo and colleagues (1998) found no significant differences in the cardiotocographic patterns of the fetus before and after MRI. In addition, no trend toward an increase or a decrease in fetal heart rate or movement was found.

Several longitudinal studies have been done to assess postnatal effects of prenatal imaging. Myers and colleagues (1998) studied 74 women who were imaged during the second and third trimesters of pregnancy and found no significant effect of imaging on premature delivery or evidence for intrauterine growth restriction. In a study on 20 infants, each imaged four times in utero, Clements and colleagues (2000) found no neurological abnormalities upon examination of the infants at nine months of age. More invasive studies have been performed on animals, High and colleagues (2000) found that a 10-week exposure of rats to a 9.4 T static magnetic field had no significant effects on spatial memory, incidence of death, changes in heart rate, body weight, food and water consumption, blood and urine composition, and physical pathology. They concluded that there were no adverse biological effects of exposure in male and female rats, as well as their progeny.

Radio Frequency (RF) electromagnetic field (B1)

The main safety risks from the radio frequency field used in MRI are tissue heating and burns.

The best method to counter this risk is a comprehensive screening of all subjects to exclude subjects with metal objects from the scanner room - either external objects such as jewelry, or internal objects such as implanted metal devices. The investigator must also make sure that conductive materials such as wire leads that might act as a RF antenna do not come in contact with the subject's skin. The MRI operator must also be certain that the pulse sequence used does not exceed FDA limits on specific absorption rate (SAR).

Thermal effects of RF for the body

Brief applications of radio frequency electromagnetic fields are used during imaging to excite the spins and tilt them from their alignment with the static magnetic field. The absorbed energy from these radio frequency fields are later emitted by the spins at the same frequency, and these emitted radio frequency waves are the signals that we sample to create MR images. However, the coil transmits more radio frequency energy during excitation than is later emitted by the spins. This excess energy can penetrate the superficial surfaces of the body and become absorbed by the body's tissues. It is dissipated in the form of heat - either through convection, conduction, radiation, or evaporation. Thus, a potential concern in MRI is the heating of the body during imaging. Shellock(2000) has provided an excellent recent review of RF heating in MRI.

The Specific Absorption Rate (SAR) is the measure of the absorption of electromagnetic energy in the body (typically in watts per kilogram, or W/kg). The rate of absorption of RF energy depends upon the RF frequency and the size, geometry and conductivity of the absorbing object. The frequency of the RF used in imaging is dependent upon the static field strength (for H1, Radio Frequency = 42.58 MHz * Field Strength in Tesla). A frequency of 63.9 MHz is used to excite H1 spins at 1.5 Tesla, and a frequency of 170.3 MHz is used at 4 Tesla. These frequencies are in the VHF or television range. Longer RF pulses (180) deposit more RF energy than shorter pulses (90), and SAR is greater for pulse sequences that employ many RF pulses per unit time (such as fast spin echo) than those that employ fewer (such as gradient echo EPI).

The SAR of imaging is limited to cause less than a one-degree (C) temperature rise in core temperature. Higher frequencies are more energetic than lower frequencies, so there is a greater potential for heating at higher static field strengths. The original FDA MRDD (Magnetic Resonance Diagnostic Device) Guidelines stated a level of concern related to whole body RF heating of 0.4 W/kg. It was later found that operation at up to 4 W/kg was possible without incurring a core temperature rise of one-degree C. As a result, MRDDs have been cleared for market operating at up to 4 W/kg since reclassification (U.S. Department Of Health and Human Services, 1998). Operation above this level requires an approved human studies protocol under the IEC standard. Experimental studies by Shellock and colleagues (1994) exposed healthy subjects to a SAR of 6 W/kg while measuring a variety of physiological indices and temperatures. All measures stayed within physiological safety limits, indicating that the current standards have a considerable margin of safety for healthy individuals.

To avoid overheating any local area, the product of time and local SAR should not exceed (in Watts per minute per kilogram):

provided that the instantaneous SAR does not exceed:

Actually determining SAR is difficult, and depends upon various models of heat conduction and body geometry. The calculation depends on the subject's weight and thus it is important to enter the subjects weight as correctly as possible. Subjects regulate heat dissipation through perspiration and blood flow changes, so it is important that the subject be made comfortable (bore fan, blanket if requested) and the subject be asked frequently how they are doing. For exposures to infants or pregnant women, a reduction of those values by a factor two is recommended by the FDA. Thermoregulation is impaired in patients with cardiocirculatory impairments, cerebral vascular impairment, and diabetes, and thus SAR should also be lowered. The FDA recommends 1.5 W/Kg for all such compromised patients.

SAR limits are enforced by software routines within the pulse sequence program (PSD) and by power monitors on the scanners that limit the duty cycle of the RF waveform delivered to the coils by the PSD.

Thermal effects of RF for foreign metal bodies

Metal devices also absorb RF energy and make become hotter than the surrounding tissue. Notably, Shellock(2000) reports that there have been no reports of burns associated with internal medical devices. However, this potential exists at high field strengths.

Certain looped configurations, particularly from wire leads, can act as a RF antenna, and thus focus RF energy to a small locus. Thus, the most significant safety risk for the RF electromagnetic fields used in MRI is a local burn (note that induced currents in conductors and loops due to time-varying magnetic fields associated with gradient coils can also result in heating).

These incidents reported to the FDA illustrate potential problems.

The potential for burns is greatest when the patient is uncommunicative, sedated, or otherwise compromised. The best methods for avoiding burns are (1) screening subjects to exclude those that have metal devices or wires implanted within their bodies, (2) ensuring that subjects remove all metal prior to entering the scanner including non-ferromagnetic jewelry such as necklaces, piercings, and earrings, (3) make sure that any wire leads (such as ECG, EMG, or EEG leads) are not looped and that wires are not run over bare skin.

Non-thermal effects of RF for implanted medical devices

In addition to heating, RF fields could interact with the operation of electronic medical devices, causing them to malfunction. For example, RF fields could interfere with the electronics of a pacemaker causing asynchronous or rapid pacing. Time-varying magnetic gradients can also interfere with the operation of implanted medical devices. Most deaths associated with MRI have resulted from scanning patients with pacemaker devices. It is absolutely essential that all potential subjects are thoroughly screened to exclude those with implanted medical devices (pacemakers, cochlear implants, infusion pumps, neurostimulators, etc).

Time varying magnet field (dB/dt)

The main safety risks from time varying magnetic fields is the generation of currents that may cause peripheral nerve stimulation and may disrupt the operation of implanted medical devices.

The best method to counter these risks is a comprehensive screening of all potential subjects to make sure that those with implanted medical devices are identified and excluded from participation. Subjects should not clasp their hands together or create other such loops with their hands. Subjects should be told to report any tingling, muscle twitches, or other sensations that may be the result of peripheral nerve stimulation.

Gradient magnetic fields are briefly imposed during image acquisition in order to spatially encode the spins. The gradient magnetic fields are much weaker than the static magnet field and are typically measured in thousandths of Tesla (mT) per meter (m). For example, both 3.0 Tesla BIAC scanners have gradients of 40 mT/m. The gradient coils are charged with current over a very brief interval to produce these gradient fields. Thus, the change in magnetic field (dB) occurs over time (dt) and is usually measured in units of dB/dt. During the rise time of the magnetic field, an electric current may be induced in a conductor. Since the human body is a conductor, currents induced by dB/dt have the potential to stimulate nerves and muscles. Currents can also be induced in implanted medical devices or wires. Currents generated in the body, or in devices implanted in the body, pose safety risks during scanning, and each possibility will be reviewed briefly below. We direct interested readers to a recent comprehensive review of patient safety in time-varying gradient fields by Schaefer and colleagues (2000).

It is important to note that dB/dt is a function of the gradient strength and rise time, and not of the static field strength. Thus, dB/dt is of equal concern for both lower and higher field scanners.

Nerve stimulation

Currents induced in the body by dB/dt can cause peripheral nerve or muscle stimulation. This stimulation may result in a slight tingling sensation or a brief muscle twitch that may startle the subject, but is not recognized as a significant health risk. Threshold sensations such as these should not be ignored, however, because this sensation may escalate to unpleasant or painful at higher levels of dB/dt. Since the margin between barely perceptible and unpleasant has been reported to be on the order of 1.5 (Schaefer et al., 2000), it is important that subjects report any sensation during scanning so that corrective action can be taken.

The FDA has altered its guidelines for dB/dt based upon recent research. In their 1998 document (U. S. Food and Drug Administration, 1998), the FDA stated "The original MRDD Guidance had established a level of concern for dB/dt at 20 T/sec for pulse duration over 120 microseconds. As an alternative, a manufacturer could demonstrate that the rate of change of the gradient field was not sufficient to cause peripheral nerve stimulation by an adequate margin of safety. The development of echoplanar and similar fast imaging techniques, and the clinical benefits which they provide, caused a re-evaluation of this policy. Evidence was presented that although peripheral nerve stimulation could potentially startle a patient and cause motion which could interfere with image acquisition, the sensation is not harmful. However, painful stimulation should be avoided." Thus, the current FDA standard is based upon the threshold for sensation, rather than a specific numerical value. With regard to dB/dt, values below that resulting in painful stimulation are considered non-significant risk by the FDA (1997). This reflects, in part, the difficulty in calculating the distribution of current with the body, a process that relies upon elaborate modeling.

The best means to address the discomfort of peripheral nerve stimulation is to instruct subjects not to clasp their hands together during scanning, and this causes a conductive loop that may potentiate dB/dt effects. Subjects should also be instructed to report any tingling, muscle twitching, or painful sensations that might occur during scanning.

It is theoretically possible to induce currents of sufficient to influence cardiac function and, in the extreme, cause ventricular defibrillation. Reilly (Reilly, 1989, 1991, 1992, 1993) investigated the dB/dt necessary to produce cardiac stimulation. His studies indicate that there is a large margin of safety between the dB/dt levels necessary for perceptible nerve stimulation and necessary to cause cardiac stimulation for ramp durations below 1000 μsec. In examining the experimental literature obtained in studies of dogs and the simulation curves of Reilly, Schaefer and colleagues (2000) state that the cardiac stimulation threshold for the most sensitive 1% of the population should require 20 times the energy required for peripheral stimulation. Furthermore, the mean defibrillation threshold should require 500 times the energy required for peripheral stimulation. For subjects experiencing a dB/dT sufficient for peripheral nerve stimulation at a 100 μsec ramp duration, the probability of cardiac stimulation is only 2 x 10-29.

Implanted medical devices

As reviewed in the section on RF fields, interference with an implanted medical device may result from both RF fields and by induction of currents within the device by dB/dt. Also, currents may be induced within implanted control wires, even if the device has been removed.

These examples were reported to the FDA.

The best means to address these risks are the same as for avoiding the risks associated with RF fields, carefully screen subjects and exclude those with implanted medical devices.


The main safety risk from the confined space of the magnet bore is the subject's physical and psychological discomfort.

The best method to counter this risk is to carefully prepare your subjects. Exclude those from participation who are known to be claustrophobic. Provide your subject with an emergency button to signal their discomfort, and communicate frequently with them between scans. Fresh air blowing through the tube can sometimes alleviate fears of suffocation.

Most subjects find the physical confinement of the MRI bore at least somewhat uncomfortable. However, for some subjects, this confinement results in anxiety and, in the extreme, panic. Studies by Harris and colleagues (1999) suggest that claustrophobic reactions in the scanner are composed of two independent components: a fear of suffocation, and a fear that something will happen while they are confined. This two-part conceptualization is similar to that of Shafran et al. (1993), who also emphasized that claustrophobic individuals believe that they have lost control while confined. The number of subjects who experience claustrophobic reactions during MRI is uncertain. Kilborn and Labbe (1990) estimated that from 5-10% of patients become claustrophobic during clinical MRI scans. Murphy and colleagues (1997) reported that 14.3% of 949 patients undergoing MRI testing in their hospital required sedation to tolerate the procedure. In a prospective study, McIsaac and her colleagues (1998) assessed 80 first time MRI patients with a psychological battery designed to evaluate anxiety. Eleven of the patients reported "panic" during the scans (undefined), and three of those patients terminated their scans prematurely. On a one-month follow-up, 30% of the patients reported that their single experience with the MRI had increased their overall level of claustrophobia.

The numbers reported in these studies are not representative of our experience at BIAC where terminated scans and panic appear to be much less frequent. This is not surprising, given that our volunteers choose to participate while patients have less choice. Also, patients may have additional anxieties about their illness or diagnosis that contribute to their discomfort. Nevertheless, claustrophobia is a significant problem that should not be ignored.

There is no simple solution to the problem of claustrophobia. One approach is to exclude from study those subjects who state that they are claustrophobic on the BIAC screening form. However, potential subjects who know they are claustrophobic do not typically volunteer for MRI studies. Experience suggests that anxiety in the scanner can be reduced by talking to subjects frequently throughout the scan - particularly at its onset, by keeping the bore fan running to reduce heat and eliminate any fear of suffocation, and by providing the subject with an emergency panic button so that he or she knows that help can be immediately summoned and that they have not lost control. For first time subjects, an experimenter should explain that the sounds they will hear are a normal part of scanning. Subjects should also be told that mild apprehension in enclosed spaces is a normal reaction, but if they feel increasingly anxious during the scan, they can ask to stop the scan. An experimenter must listen for telltale signs of growing anxiety or discomfort, such as the subject repeatedly asking how much longer the scan will last. Taking a few minutes to enter the scanner room and reassure a subject may help avoid an escalation of anxiety. However, if a subject appears to be more than mildly anxious or declares himself or herself to be significantly anxious, then the experimenter must remove the subject from the scanner immediately.

Acoustic noise

The rapid changes of current with the gradient coils cause the coils to vibrate due to Lorentz forces. Loud noises are created when these moving coils and their mountings vibrate.

Hearing protection - earplugs or earplugs + headphones must be worn by all subjects, and by all accompanying persons who are in the magnet room during scanning. Make sure that the subject has properly inserted the earplug.

Exposure limits for acoustic noise

The Occupational Safety and Health Administration (OSHA) regulates noise exposure in most workplaces. OSHA's noise regulations are based upon a time-weighted average (TWA) of noise exposure limited to 90 dB (A) averaged over an 8-hour day. Thus, the OSHA standard recognizes that the risks associated with acoustic noise is a function of its intensity and the duration of exposure. For purposes of the hearing conservation program, noise exposures for employees are without regard to any attenuation provided by the use of hearing protection devices, such as earplugs or earmuffs (headphones).

Continuous1 noise in db (A) measured on slow response OSHA2 maximum exposure per day (hrs)
85 16
90 8
95 4
100 2
105 1
110 0.5
115 0.25
1If the variations in noise level involve maxima at intervals ≤ 1 sec, it is considered continuous.
2Adapted from Table G-16a OSHA Regulations for Noise Exposure 1910.95

Remember, the decibel scale is logarithmic, thus a SPL that is 10 dB higher than a reference sound is 10 times louder. A 6 dB difference represents a doubling of sound intensity. To provide some real-world references, a normal conversation is typically 60 dB, a Walkman can reach 100 dB, and the sounds at the front row seats at a rock concert have been measured at 110 - 120 dB. The threshold of pain is about 130 dB.

Exposure to impulsive or impact noise must not exceed 140 dB peak sound pressure level. The impulsive noise exposure limit of 140 dB peak of the 1972 OSHA regulation does not specify a limit for the number of impulses that a person can be exposed to in an 8-hour working day.

The FDA indicates that acoustic noise in a MRI system must be below the level of concern established by the appropriate federal regulatory or other standards-setting organization (cited in McJury and Shellock, 2000). It is worth noting that OSHA does not regulate non-occupational exposure to noise, such as that experienced by a subject or patient in a MRI scan (McJury and Shellock, 2000). In its documentation for premarket notification for MRI systems (1998), the FDA states, "A magnetic resonance diagnostic device should not produce noise having an unweighted peak sound pressure level higher than 140 dB". This is essentially the maximum allowed for impulse noise (see above). Stricter noise standards have been proposed by the IEC that will limit average noise in the MRI scanner to 99 dB, but these have not yet been adopted by the FDA.

Sources of acoustic noise

When the scanner is not acquiring images, the only source of acoustic noise is the chirping of the pump used to circulate cold water. Different studies have report this noise to be in the 65-80 dB (A) range, below the OSHA standard (see table above) necessitating the use of a hearing protection device (HPD).

The time-varying gradient magnetic fields are the primary source of acoustic noise during MR imaging. The rapid rise and fall of currents within the gradient coils in the presence of the static magnetic field create strong Lorentz forces that cause the gradient coils to move against their mountings. The vibration of the coils and the vibration and flexing of their mountings cause the loud tapping and knocking noises during imaging. An interesting java applet demonstrating Lorentz forces can be found at this website.

There are many factors that influence the intensity of the acoustic noise including gradient strength and slew rate. Price and colleagues (2001) parametrically varied imaging parameters on one system and observed a 7.6 dB (A) range of values depending upon sequence. Higher sound pressure levels (SPLs) were associated with smaller FOVs and shorter TRs. Echoplanar (EPI) and other fast image sequences are typically louder than conventional sequences, particularly in the high frequency range. Miyati and colleagues (1999) compared acoustic noise profiles of echoplanar (EPI) and other sequences on eleven MRI systems including four GE 1.5 T Horizons with 18.8 mT/m gradients. Across all systems, the mean A-weighted SPL (Leq) was 94.1 dB and Lpeak was 108.9 dB for the EPI sequences. The greatest acoustic energy was recorded in the 1000-4000 Hz range. EPI did not differ significantly from other fast pulse sequences (such as FSE or SPGR) overall, but there was more energy for single-shot EPI at 4000 Hz. Interestingly, among the four identical model GE systems, there was a 7.0 dB difference in Leq and a 7.7 dB difference in Lpeak while running the same EPI sequence. This suggests that the room setup and other site-specific variables can contribute significantly to the noise levels. It has been suggested that a significant proportion of the gradient noise may be reflected from the walls and could be reduced with sound absorbent materials.

Miyati and colleagues (1999) tested 0.5 T, 1.0 T and 1.5 T scanners and found no significant differences in SPL due to field strength, although the higher field magnets tended to be somewhat noisier. Price and colleagues (2001) did find a significant difference in acoustic noise related to field strength in a comparison of 15 MRI systems varying in field strength from 0.2 T to 3 T. However, the field strength comparison was confounded with differences in gradient strength (i.e., the 0.2 T system had 10 mT/m gradients and the 3 T had 35mT/m gradients). For the two 3T systems tested (Varian and Bruker), the SPLs averaged 115 dB (A). It is notable that the highest SPLs were not recorded at isocenter, but rather about 80 cm away at the entrance to the bore. Sound intensity within the scanner was influenced by the presence of a subject. Higher SPLs were recorded with a subject present than without.

Note: This paragraph refers to retired equipment. The BIAC 1.5 T and 4.0 T scanners were tested for sound intensity during image acquisition using echoplanar imaging with spin echo (EPI-SE) and gradient echo (Duke EPI). For the 1.5 T scanner at magnet isocenter, the Leq was 93 dB (A) for EPI-SE and 98 dB (A) for Duke EPI. For the 4.0 T scanner at magnetic isocenter, the Leq was 93.9 dB (A) for EPI-SE and 98.2 dB (A) for Duke EPI. Thus, the scanners fall within OSHA guidelines for occupational exposure for the typical length of a MRI session.

Sequences involving high diffusion weightings, which utilize the full gradient range, are likely to produce louder sounds. At present, no SPL measurements have been performed using these sequences.

Hearing protection devices

The EPA has established a standard — the Noise Reduction Rating or NRR — for hearing protection devices. The NRR value is determined by determining the mean attenuation values (in dB) of the hearing protection device and then subtracting two standard deviations. Thus, the NRR should represent the minimum noise reduction achieved by 98% of the population.

The Aearo E-A-R Classic earplugs used at BIAC have an EPA Noise Reduction Rate (NRR) of 29 dB. However, the rated NRR of an earplug may not be met if they do not fit properly or if they have been improperly inserted. Indeed, for meeting compliance, NIOSH derates earplug NRR by 50% if no subject fit data is available. You must instruct the subject on the proper insertion procedure for the earplug and then inspect the fit to make certain that this procedure has been followed. Instructions on properly fitting Aearo earplugs can be found at the E-A-R website. Instructions can also be found in the MRI suite.

Wearing headphones and earplugs together do not produce additive effects in hearing protection, but can increase the NRR by an additional 5-6 dB (equivalent to approximately half the sound intensity). At >2000 Hz, hearing protection is bone conduction limited and the combination of headphones and earplugs is least effective. However, at lower frequencies there is considerable gain in using both devices in combination. As there is considerable low frequency sound in MRI, the use of earplugs and headphones should be considered.

Safety Video and Quiz

If you need access to the MRI suite, you should view the MRI Magnet Safety Video (requires Windows Media Player) and take the MRI Safety Quiz to demonstrate your understanding of the material covered in this safety lecture. See SOP111: BIAC Safety Training for more information.

Additional resources


Chaljub G, Kramer LA, Johnson RF, III, Johnson RF, Jr, Singh H, Crow WN (2001) Projectile cylinder accidents resulting from the presence of ferromagnetic nitrous oxide or oxygen tanks in the MR suite. American Journal of Roentgenology 177:27-30.

Clements H, Duncan KR, Fielding K, Gowland PA, Johnson IR, Baker PN (2000) Infants exposed to MRI in utero have a normal paediatric assessment at 9 months of age. British Journal of Radiology 73:190-194.

Harris LM, Robinson J, Menzies RG (1999) Evidence for fear of restriction and fear of suffocation as components of claustrophobia. Behaviour Research & Therapy 37:155-159.

High WB, Sikora J, Ugurbil K, Garwood M (2000) Subchronic in vivo effects of a high static magnetic field (9.4 T) in rats. Journal of Magnetic Resonance Imaging 12:122-139.

Kanal E, Gillen J, Evans JA, Savitz DA, Shellock FG (1993) Survey of reproductive health among female MR workers. Radiology 187:395-399.

Kilborn LC, Labbe EE (1990) Magnetic resonance imaging scanning procedures: development of phobic response during scan and at one-month follow-up. Journal of Behavioral Medicine 13:391-401.

McIsaac HK, Thordarson DS, Shafran R, Rachman S, Poole G (1998) Claustrophobia and the magnetic resonance imaging procedure. Journal of Behavioral Medicine 21:255-268.

McJury M, Shellock FG (2000) Auditory noise associated with MR procedures: a review. Journal of Magnetic Resonance Imaging 12:37-45.

Miyati T, Banno T, Fujita H, Mase M, Narita H, Imazawa M, Ohba S (1999) Acoustic noise analysis in echo planar imaging: multicenter trial and comparison with other pulse sequences. IEEE Transactions on Medical Imaging 18:733-736.

Murphy KJ, Brunberg JA (1997) Adult claustrophobia, anxiety and sedation in MRI. Magnetic Resonance Imaging 15:51-54.

Myers C, Duncan KR, Gowland PA, Johnson IR, Baker PN (1998) Failure to detect intrauterine growth restriction following in utero exposure to MRI. British Journal of Radiology 71:549-551.

Poutamo J, Partanen K, Vanninen R, Vainio P, Kirkinen P (1998) MRI does not change fetal cardiotocographic parameters. Prenatal Diagnosis 18:1149-1154.

Price DL, De Wilde JP, Papadaki AM, Curran JS, Kitney RI (2001) Investigation of acoustic noise on 15 MRI scanners from 0.2 T to 3 T. Journal of Magnetic Resonance Imaging 13:288-293.

Reilly JP (1989) Peripheral nerve stimulation by induced electric currents: exposure to time-varying magnetic fields. Medical & Biological Engineering & Computing 27:101-110.

Reilly JP (1991) Magnetic field excitation of peripheral nerves and the heart: a comparison of thresholds. Medical & Biological Engineering & Computing 29:571-579.

Reilly JP (1992) Principles of nerve and heart excitation by time-varying magnetic fields. Annals of the New York Academy of Sciences 649:96-117.

Reilly JP (1993) Safety considerations concerning the minimum threshold for magnetic excitation of the heart. Medical & Biological Engineering & Computing 31:651-654.

Rodegerdts EA, Gronewaller EF, Kehlbach R, Roth P, Wiskirchen J, Gebert R, Claussen CD, Duda SH (2000) In vitro evaluation of teratogenic effects by time-varying MR gradient fields on fetal human fibroblasts. Journal of Magnetic Resonance Imaging 12:150-156.

Schaefer DJ, Bourland JD, Nyenhuis JA (2000) Review of patient safety in time-varying gradient fields. Journal of Magnetic Resonance Imaging 12:20-29.

Schenck JF (2000) Safety of strong, static magnetic fields. Journal of Magnetic Resonance Imaging 12:2-19.

Shafran R, Booth R, Rachman S (1993) The reduction of claustrophobia--II: Cognitive analyses. Behaviour Research & Therapy 31:75-85.

Shellock FG (2000) Radiofrequency energy-induced heating during MR procedures: a review. Journal of Magnetic Resonance Imaging 12:30-36.

Shellock FG, Schreyer AG, Kanal E (1994) Physiological responses to MR imaging performed at a SAR level of 6.0 W/Kg. Radiology 192:865-869.

U. S. Food and Drug Administration (1997) Magnetic resonance diagnostic devices criteria for significant risk investigations. In:

U. S. Food and Drug Administration CDRH (1998) Guidance for the Submission Of Premarket Notifications for Magnetic Resonance Diagnostic Devices.

Wiskirchen J, Groenewaeller EF, Kehlbach R, Heinzelmann F, Wittau M, H.P. R, Claussen CD, Duda SH (1999) Long-term effects of repetitive exposure to a static magnetic field (1.5 T) on proliferation of human fetal lung fibroblasts. Magnetic Resonance in Medicine 41:464-468.