Operational safety issues in MRI

Copyright 8

0730-7255187 F3.00 4 00 19X7 Pcrgamon Journal\ Lid.

l Original Contribution

OPERATIONAL RAYMOND

E.

GANGAROSA,

DUANE

PRASCHAN,

SAFETY

ISSUES IN MRI

JANET E. MINNIS, AND RICHARD

W.

JANINE NOBBE, GENBERG

Picker International,Clinical ScienceCenter, 5500 Avion Park Drive, Highland Heights, OH 44143 The clinical usefulness of a diagnostic modality is weighed by considering its potential diagnostic benefit against its potential risk for a patient in question. Magnetic resonance imagining appears to offer both high efficacy and safety under most circumstances. Our understanding of the conditions under which MRI is safe and effective has undergone continual refinement with technological advances and clinical experience. The early emphasis on safety issues of MR focussed on consideration of bioeffects of RF and magnetic fields. More recently, hundreds of operating clinical MR sites, performing hundreds of thousands of clinical examinations to date, have provided a greater awareness of operational safety issues. Much of this experience is summarized in device labeling provided by manufacturers of MR devices, summaries prepared by regulatory agencies, and case reports in the medical literature. The purpose of this article is to review a broad range of safety considerations involved in the operation of MR imagers. The discussion is in two parts: (1) a short update of reported incidents and (2) an analysis of safety issues. Keywords: MRI safety; MRI operational

REPORTED

safety; MRI hazards; Magnetic resonance safety.

patient with a pacemaker was scanned. The patient arrested during the scan; it was later reported that the patient’s EEG was flat. No further details are available. A patient with a neurostimulator who was scanned indicated that he felt pain when the pulsing of the imager began. The imager was determined to be the cause of the pain. A patient with a previously undiscovered orbital iron filing had a subretinal hemorrhage with subsequent unilateral visual loss (see below). Industrial accidents have been reported in which workers have had traumatic injuries caused by forceful attraction of ferromagnetic objects (see below). A couple incidents of eye or periorbital irritation following MR examination, caused by eye makeup containing ferromagnetic ingredients, have come to our attention through discussions with the FDA. One incident involved externally applied eye makeup; the force of the magnet was believed to displace the makeup into the eye, causing irritation. Another incident involved cosmetic eye tattooing; localized swelling occurred four hours after the examination. A patient complained of sudden unilateral ear pain,

INCIDENTS

We have recently learned of a number of incidents that have occurred during MR imaging, as reported throughout the U.S. from all manufacturers. These reports were obtained from pooled FDA summaries of medical devices,’ accounts published in the medical literature, and discussions with the FDA. Due to the sketchy information available to us, we will not comment on them in detail, nor draw explicit connections with the analysis, of safety issues. One death occurred during MR examination. The only reported information was that the patient expired while undergoing an examination for cerebral infarction. Two cardiac arrests have occurred during MR examination. One was a child with a cerebellar tumor who was on mechanical ventilation and had a history of prior cardiac arrest. It was reported that cardiac monitoring leads interfered with the scan, and were removed, while the patient was visually observed during the scan. At the end of the scan, the child went into cardiac arrest. The child was resuscitated. The other cardiac arrest was reported when a RECEIVED 3/23/87;

Address

ACCEPTED 4/3/87. 287

correspondence

to Dr. Raymond

E. Gangarosa.

Magnetic

288

Resonance

imaging 0 Volume

and later complained of dried blood in that ear and problems with nerves in some teeth. Two patients have complained of persistent hearing problems (ringing in the ears, “ocean sounds”). To our knowledge, these reports did not attribute the problems to the MR examination. . Miscellaneous incidents: A few cases are reported in which right-left reversal of images occurred due to miswiring errors. A few cases of malfunction OCcurred in patient handling systems or intercom. One patient had a minor hand laceration while being moved into the magnet. SAFETY

ISSUES

Electrically, Magnetically, and Mechanically Activated Implants The FDA has specifically alerted the MR community to the dangers of implanted pacemakers and neurostimulators; patients with such prostheses should be excluded from the vicinity of the imager. Early experiments demonstrated that at static fields as low as 10 gauss, some pacemakers switch modes (typically from demand to asynchronous) due to magnetic attraction of reed relay components within the pacemaker. There is concern that at yet higher fields, typical of imaging exposures, such relays might be permanently damaged. Also at close range, the rapidly switched audio- and radio-frequency magnetic signals used in imaging might be coupled into the circuitry of the prosthesis and superimposed on the prosthesis’ electronic waveforms, with potentially disastrous results. Other prostheses in this category might include cochlear implants (electronic) and stapedial replacement prostheses (mechanical). Some evidence for electromagnetic interference in cochlear implants has been cited in other contexts.’ Stapedial implants may represent a relative contraindication to MR examination unless they are known not to be ferromagnetic. We have not determined which stapedial implants are ferromagnetic, but in the absence of additional information all patients with such prostheses should be excluded. It seems likely that many such prostheses could be designed to be compatible with the MR environment. However, such examinations should be performed in specifically-designated research collaboration with the manufacturer. FERROMAGNETIC

PROSTHESES

Considerable experience has been accumulated for MR imaging of patients with weight-bearing and wellheated surgical prostheses, with no reported adverse

5, Number

4, 1987

effects. Because of the risk of hemorrhage, clips for surgical hemostasis are of greatest concern. In most body regions, healing of hemostasis clips is accompanied by fibrosis and encasement of the clip. However, this process does not occur in the CNS, so affected surgical clips, particularly aneurysm clips, are contraindicated. For patients with ferromagnetic clips in other body regions, sufficient time should be allowed postoperatively for fibrosis to anchor the clips. The required time interval might vary. The required time interval might be variable, for example if healing is poor, as for areas of poor blood supply, extremely sick patients, etc. Since the force on a ferromagnetic object may increase with the square of the magnetic field, policies should be especially conservative for high field imagers. A better solution to the problem is to use only clips that are not magnetically active. Ex vivo testing of a sample of the same implant (even through sterile packaging) with an ordinary bar magnet may offer some assurance, and might suffice for examinations at low field. However, the attraction force on a magnetic material can change with work hardening, lot-to-lot variability, and other confounding issues. The best policy is to ensure that the surgical department uses only implants and clips certified by the manufacturer to be nonmagnetic. Even when this policy is instituted, it is important to recognize that clips implanted previously or elsewhere may still present hazards. An even more difficult problem is that of the occult ferromagnetic foreign body. One serious accident involved a former sheet metal worker who had a subretinal hemorrhage and subsequent unilateral visual loss from a previously undetected metal iron filing.*3 No satisfactory screening method currently exists to identify such patients at risk; history and x-ray examination would not generally be sufficiently sensitive and specific for detection of ferromagnetic objects.? One possible, though not fully satisfactory, method for reducing this risk would be to move the patient into and out of the magnet slowly, and to stop if the patient notes any adverse symptoms (tugging sensation and flash of light were reported in this case). The problem with ferromagnetic objects would be considerably reduced if a simple, inexpensive, and rapid test could determine whether a known or presumed implant is ferromagnetic. X-ray examination *This incident occurred at relatively low field strength (0.35 T). -iThe patient described in the reference above did not know he had an iron filing in his orbit, although he admitted to performing sheet metal work without using eye protection.

Operational safety issues in MRI 0 R. E.

may be useful in some cases, eg., to detect whether an aneurysm clip is present, but cannot determine the magnetic property of a metal object. Finn et aL4 have evaluated a number of commercially-available devices which might be useful for screening for magnetic materials, but at present this is an unfilled need. Early workers5 suggested caution in exposure of patients with prosthetic heart valves to MRI. More recently, Soulen et aL6 have investigated ex vivo magnetic forces and temperature effects on nine different synthetic and tissue valves in whole body imagers of 0.35 and 1.5 T and a smallbore electromagnet of nominal 2.35 T, with peak RF power of 800 W and measured static field gradients of up to 6.3 mT/cm. The valves tested were all of recent design using primarily nonferromagnetic materials. Using available pulse sequences, no temperature rise could be detected in any of the prosthetic valves. For the valves tested, no deflection was measured at 0.35 T, although small deflections were measured at 1.5 T and significant deflections in the smallbore system in some valves. The magnetic forces were felt to be small compared to those exerted by the heart; however, for some older valves, the authors recommend limiting exposure to the lower field if valve dehiscence is clinically suspected. They also report that, although artifacts were present locally around the prosthetic valves, they did not preclude diagnostic examinations, but that safety and diagnostic usefulness would have to be reevaluated if exposure conditions were increased further. PHYSIOLOGICAL

MONITORING

Here we will define physiological monitoring devices as instrumentation used in life support by continuous recording of vital waveforms. Many of the problems described above for electronic prostheses during MRI also apply to certain physiological monitors. In a manner similar to radio reception, signals from the imager’s gradient amplifiers and RF transmitter can be coupled into electronic monitoring circuitry. This might then be amplified to create confusing artifacts superimposed on the physiological waveform. ECG monitoring circuitry performs RF filtering, but the gradient waveforms contain (audio-frequency) components within the frequency band of the ECG signal. To our knowledge, no work has been done to develop, test, and market equipment suitable for cardiac monitoring during MRI. ECG signals obtained during monitoring of patients in high magnetic fields will typically demonstrate pronounced T-waves, due to the magnetohydrodynamic effect of flowing blood.’ This effect results from the voltage induced by the flow of conductive

GANGAROSA, ET AL.

289

blood in a large static magnetic field, and produces no physiologic hazard; however, it must be taken into account when evaluating ECGs of patients in mid- to high-field imagers. The timing of the spurious peak will depend on the timing of systolic ejection, and the amplitude on peak cardiac output, as well as orientation of the body and electrodes with respect to the static field. It is possible that the flow-induced artifact may provide additional diagnostic information from ECGs obtained during MRI examinations. One simple approach to the problem of cardiac monitoring during scanning would be to use a nonelectronic transducer, such as a plethysmographic or laser doppler pulse monitor. Such devices have been described for use in cardiac gated MR1.s We are not aware of any efforts to make them available as physiological monitors. Engler et aL9 have shown that many intravenous infusion pumps are affected by the MR imager. As exposure conditions may vary, any prospective device should be tested on the imager for which use is intended. Dunn et al.” have found one solution to the problem of mechanical respiration during MR imaging. The respirator they used$ works on the principle of fluidics, whereby switching is performed through pressure applied to jets of air. Since electronics are not used, no spurious signals are picked up. The device also is not affected by magnetic field at 0.5 T, and can be used in close proximity to the magnet bore. This group reports highly satisfactory results at 0.5 T in a large group of patients that would not otherwise be candidates for MRI. ATTRACTION OF FERROMAGNETIC OBJECTS Forceful attraction of ferromagnetic objects represents the clearest danger to the patient during MRI. The force on ferromagnetic objects may increase as strongly as the square of the magnetic field, so the danger is particularly great at mid- and (especially) high-field. The force of the magnet can propel a ferromagnetic object alternately through both sides of the imager in an oscillatory manner until it hits something or slowly comes to rest in the center of the field. Although no accidents have involved patients, a serious, lifethreatening accident during assembly has been reported. ’ ’ It is vitally important to recognize that forces on ferromagnetic objects are deceptively nonlinear with decreasing distance. At the edge of the imaging suite, SMonaghan 225 SIMV Volume Ventilator.

290

Magnetic

Resonance

Imaging

even a highly ferromagnetic object will typically exhibit no significant attraction to the imager. As the object is moved closer, it may be suddenly pulled into the imager. A highly ferromagnetic object may become violently and uncontrollably wrenched from the hands of a person carrying it. To indicate a zone of potential danger, we currently recommend marking a 50 gauss line on the floor around the imager. We are not aware of fully-tested, approved, practical screening devices for discriminating ferromagnetic objects; thus, safety relies on surveillance and awareness. The operator must be especially alert to the entry of any personnel into the imager suite while anyone is in the magnet. Persons unfamiliar with the imager must be warned to remove ferromagnetic objects, and, if possible, to defer entry until no one is in the imager. It may be helpful to install (securely!) small magnets well outside the imager suite, which can be used to test any objects that might be ferromagnetic. If the suspected object is attracted to the test magnet, it should be kept outside the imaging suite. EMERGENCY

PLANS

In the report cited above,” emergency personnel encountered minor problems with the magnetic field. Drills to acquaint emergency personnel of rescue procedures may obviate potential hazards. RF HEATING RF energy can produce a heating effect similar to that used in short wave diathermy (at 27 MHz) to promote blood flow. Excessive temperature rise caused by intense RF heating could result in tissue damage. Since it is difficult to measure internal temperature changes, we must estimate RF heating effects by absorbed power density, or specific absorption rate (SAR). Budinger’ has suggested a safe threshold for SAR of 1.5 W/Kg for long duration studies and 4 W/Kg for studies of 10 minutes or less. The FDA” has identified a guideline for “nonsignificant risk” of 0.4 W/Kg averaged over the whole body.” For comparison, the metabolic rate of humans is 1.5 W/Kg during sleep, 5 W/Kg during moderate exercise, and 15 W/Kg for very heavy exercise. Assuming no heat loss, 1 W/Kg would raise temperature by 1 degree centigrade in about 1 hour; 5 W/Kg would take 12 minutes to produce the same temperature rise. For most organs (es., skin, brain, “Another guideline was specified: 2 W/kg over any 1 gram of tissue. However, no method has been devised for measuring the worst case of tissue heating, so current efforts have centered around controlling overall heat deposition.

0 Volume

5. Number

4, 1987

liver, kidney), the rich blood supply would conduct heat rapidly from the tissue volume, considerably reducing the temperature rise. Furthermore, larger patient body mass provides a larger heat sink to dissipate the heat load. However, a few organs have much lower perfusion, such as the lens of the eye and the testes, and are more susceptible to RF heating. In addition, the FDA has expressed particular concern” for patients with circulatory compromise, eg., stroke, for whom the mechanism of heat removal is impaired. SAR increases with field strength and, for current pulse sequences, is primarily a concern only at high field strengths (1 .O T or above). SAR values obtained depend on the operating frequency, RF power, RF duty cycle, coil configuration, and body size. Any of a number of steps may be taken to alter the pulse sequence parameters to reduce SAR. These maneuvers are listed below: (1) Change the pulse sequence to one which uses fewer RF pulses per unit time, eg., a field echo instead of a spin echo, or a single echo instead of a multiecho sequence (2) Decrease the number of slices (3) Lengthen the repeat time (4) Decrease the RF scale factor (5) If practical, switch to a smaller transmitter coil Heating effects might, in principle, be increased locally in the presence of conductive objects, eg. metallic implants, due to regional distortion of RF fields. As for whole body SAR effects, higher field and higher peak RF power will accentuate‘localized heating. These effects would increase with the size and conductivity of the implant and diminish with the surrounding tissue mass and vascular perfusion (due to heat sinking). In extensive patient experience to date with prostheses of many kinds, no clinical signs of localized heating have been reported. Ex vivo temperature measurements of fairly large metal surgical prostheses under moderate exposure conditions13 revealed no measureable temperature rise. With higher RF power and field strength, the issue of localized heating may require further investigation. ACOUSTIC NOISE An MR imager produces acoustic noise much as an electric speaker: by Faraday’s law, a loop of wire (gradient coil) in a static magnetic field induces a force when a rapidly-changing electric current pulses through it. This force is transmitted to the surrounding enclosure as a vibration, which propagates as acoustic noise. Acoustic noise is accentuated at higher fields,

Operational

safety

issues in MRI 0 R. E. GANGAROSA,

higher gradient strengths, higher gradient duty cycles, and sharper pulse transitions. The latter conditions are characteristic of pulse sequences which obtain highest information content per unit time (high-speed, multiecho, and/or multislice scans), thin slices, and small fields of view. Average noise levels of 95 dBA have been measured in a Picker 1.5 T Vista@HP MR (Picker International, Highland Heights, OH) imager with peak 10 mT/m gradient strengths on all axes under worst-case switched gradient waveform (non-imaging) conditions. A continuous noise level of 95 dBA corresponds to a permissible occupational exposure of two hours per day,” I4915 and is comparable to the noise levels of very heavy traffic (92 dBA) or light road work (90-l 10 dBA). For typical imaging pulse sequences, noise levels are considerably lower; 1.5 T multislice sequences produce sound levels of 82 dBA under worst-case conditions.7 Patient exposures to these (peak) noise levels are considerably shorter than the allowed occupational exposures and are not repeated at daily intervals, as the occupational standards are considered; therefore, patient exposures are within standards allowed for healthy individuals. Patients with diseases that might make such noise exposures stressful or dangerous should, of course, be given special consideration, and pulse sequences limited accordingly (to thicker slices, larger fields of view, and conventional protocols). For greater comfort and assurance to the patient, we have often used plastic ear muffs or throw-away moldable earplugs when loud scans are performed. Ear muffs typically give 40 dB of isolation, while earplugs give about 25-30 dB. OCCUPATIONAL

291

ET AL.

mice to static magnetic fields has been reported” without identification of specific hazards. We can only advise caution during pregnancy, and particularly the first trimester, without defining specific guidelines. A magnetic field exposure meter is available,” although we are not aware under what circumstances its use would be indicated. The bioeffects of MR exposure have been widely reviewed20~2’~22~23 and are outside the scope of this article. Beryllium alloy tools, which are nonferromagnetic and have been used to reduce the projectile hazard around high field magnets, can produce incidental exposure hazards. Beryllium dust is toxic to the lungs in low concentrations; tools made of this material should only be sharpened or machined with special facilities. Hand washing is recommended after handling beryllium alloy tools. Nonferromagnetic copperaluminum-magnesium-bronze alloy or tungsten tools are preferred as nontoxic alternatives to beryllium. CONCLUDING

REMARKS

The authors hope that this review will provide a useful guide for clinical decision-making in the operation of MR centers. Ultimately, clinical judgement is the basis for determining conditions for MR exposure and/or examination. In ambiguous cases, the potential risk (documented or theoretical) must be weighed against the possible benefit of diagnostic information obtainable. In those cases, the decision may be individualized to each patient in question. Information to help guide the decision can be obtained by device labeling, ongoing experience, current knowledge, and medical literature.

EXPOSURES

One difficult question relating to occupational health is magnetic exposure of the pregnant technician. The FDA has not established a position on this issue. Teratogenic studies of magnetic field exposures are confusing and contradictory; although possible mechanisms for embryological damage have been proposed, I6 no consistent teratogenic effects have been reported. One study” comparing in vitro MR and x-ray oncogenic and genotoxic effects uncovered no deleterious effects of the MR environment. A two-year exposure study of multiple consecutive generations of “Occupational safety and health standards and interpretations, 1910.95(g) (l), pp. 143-144.14. These standards, set by OSHA, are based on lifetime occupational exposures of 35 years, and are prorated to prevent permanent hearing loss due to the cumulative effect of the exposure. nFor noise levels of 82 dBA, no occupational restrictions apply.

REFERENCES 1. Medical Device Reporting, Food and Drug Administration, Office of Compliance, selections through 1986. 2. Hepfner, S.T.; Skelly, M.F. Radio-frequency interference in cochlear implants (letter). N. Engf. J. Med. 313(6):387, 1985. 3. Kelly, W .M.; Paglen, P.G.; Pearson, J.A.; San Diego, A.G.; Solomon, M.A. Ferromagnetism of intraocular foreign body causes unilateral blindness after MR study. AJNR 7:243-245. 1986. This incident occurred at relatively low field strength (0.35 T). 4. Finn, E.J.; DiChiro, G.; Brooks, R.A.; Sato, S. Ferromagnetic materials in patients: Detection before MR imaging. Radiology 156:139-141, 1985. 5. Pavlicek, W.; Geisinger, M.; Castle, L.; Borkowski, G.P.; Meany, T.F.; Bream, B.L.; Gallagher, J.H., The effects of nuclear magnetic resonance on patients with cardiac pacemakers. Radiology 147:149-153, 1983. 6. Soulen, R.L.; Budinger, T.F.; Higgins, C.B. Magnetic resonance imaging of prosthetic heart valves. Radiology 154:705-707, 1985.

292

Magnetic Resonance Imaging 0 Volume 5, Number 4, 1987

7. Budinger, T.F. Nuclear magnetic resonance (NMR) in vivo studies: Known thresholds for health effects. J. Comp. Assisted Tomog. 5:800-81 I, 1981. 8. Lanzer, P.; Botnivick, E.H.; Schiller, N.D.; Crooks, L.E.; Arakawa, M.; Kaufman, L.; Davis, P.L.; Herfkens, R.; Lipton, M.J.; Higgins, C.B. Cardiac imaging using gated magnetic resonance. Radiology 150: I21 127, 1984. 9. Engler, MS.; Engler, M.M. The effects of magnetic resonance imaging on intravenous infusion devices. Western J. Med. 143:329-332, 1985. 10. Dunn, V.; Coffman, C.E.; McGowan, J.E.: Ehrhardt, J.C. Mechanical ventilation during magnetic resonance. Mug. Res. Imag. 3:169-172, 1985. I I. Fowler, J.R.; terpenning, B.; Syverud, S.A.; Levy, R.C. Magnetic field hazard (letter). N. Engl. J. Med. 314(23):1517, 1986. 12. Department of Health and Human Services, FDA. Guidelines for evaluating electromagnetic exposure risk for trials of clinical NMR systems: February 25, 1982. 13. Davis, P.L.; Crooks, L.; Arakawa, M.; McRee, R.; Kaufman, L.; Margulis, A.R. Potential hazards in NMR imaging: Heating effects of changing magnetic fields and RF fields- on small implants. A JR I37:857860, 1981. 14. Braver, R. Noise: Threshold of danger, Safetv Standards, Jul-Aug: 2-6, 1972. 15. Hovey, M.J. Complying with OSHA noise standards, Safety Standards, Jul-Aug: 7-12, 1972. 16. Delgado. J.M.R.; Lea], J.; Montegudo, J.L.; Garcia.

17.

18.

19.

20.

21.

22.

23.

M.G. Embryological changes induced by weak, extremely low frequency electromagnetic fields. J. Anafomy 134:857-860, 1981. Geard, C.R.; Osmak, R.S.; Hall, E.J.; Simon, H.E.; Maudsley, A.A.; Hilal, S.K. Magnetic resonance and ionizing radiation: A comparative evaluation in vitro of oncogenic and genotoxic potential. Radiology 152: 199202, 1984. Osbakken, M.; Griffith, J.; Taczanowsky, P. A gross morphologic, hematologic, and blood chemistry study of adult and neonatal mice chronically exposed to high magnetic fields. Msg. Rex Med. 3:502-517, 1986. Fujita, T.Y.; Tenforde, T.S. Portable magnetic field dosimeter with data acquisition capability. Rev. Scientific Jnstru. 53:326-331, 1982. Berhardt, J.H. (ed.). Biological effects of static and extremely low frequency magnetic fields, BGA Schriften 3/86. Institute for Strahlenhygiene des Bundesgerunstheitsamtes, MMV Medizin Verlag, Munich, 1986. Persson, B.R.R.; Stahlberg, F. Potential health hazards and safety aspects of clinical NMR examinations. Radiation Physics Department, Lasarettet, S-221 85 Lund, Sweden, 1984. Tenforde, T.S.; Budinger, T.F. Biological effects and physical safety aspects of NMR imaging and in-vivo spectroscopy. Report LBL 20053, Lawrence Berkeley Laboratory, University of California. Berkeley, CA, 1985. Shellock, F.G. Biological effects of MRI: A clean safety record so far. Diagnostic Imaging 9:96-101, 1987.