MAGNETIC RESONANCE IN MEDICINE 14,409-414
( 1990)
Sensory Stimulation by Time-Varying Magnetic Fields MARKS . COHEN, *$ ROBERTM. WEISSKOFF,* RICHARDR. RZEDZIAN, * AND HOWARD L. U N T O R ? *AdvancedNMR Systems, Inc., Woburn,Massachusetts 01801;and t Department of Cardiology, Massachusetts General Hospital, Boston, Massachusetts Received September 6, 1989; revised February 14, 1990 When two human volunteers were imaged with magnetic field gradient dB/dt of 61 T / s RMS, the subjects reported, to our surprise, feeling muscular twitches synchronous with gradient pulses over repeated experiments. No adverse or sustained effects were seen. Experiments in a canine, intended to assess the safety of MR imaging with dB/dt of up to 66 T / s RMS, failed to induce detectable changes in the electrocardiogram or to show any signs of gross response to gradient pulsing. Although these data are preliminary, and largely anecdotal, they suggest a level above which such stimulation may occur. We believe that this is the first report of direct human stimulation in an MRI device and that determination of the stimulation threshold may have impact on the selection of appropriate operating points for magnetic imaging systems. o 1990 Academic Press, Inc. INTRODUCTION
In modern whole body magnetic resonance imaging ( MRI) instruments, patients are exposed to a variety of time-varying magnetic fields. While considerable experience has demonstrated the overall safety of these devices, relatively little information is available to indicate the magnitude of the safety margin. In the course of designing our instant imaging equipment, and prior to imaging humans, we performed a series of experiments on a dog, primarily in order to assess the safety of the device. Our initial experiments failed to produce detectable effects and indicated no cause for concern. In our human experiments, however, the two subjects unexpectedly described sensations which we interpret as the effects of direct stimulation. We therefore report here what we believe to be the first description of sensory stimulation by magnetic field gradients of the type used in clinical magnetic resonance imaging systems. While these results are preliminary we feel that their importance in the safe construction of magnetic imaging systems should not be overlooked. To spatially encode the magnetic resonance signal, MR imagers produce timevarying magnetic field gradients which are a small fraction of the main magnetic field. These coil sets typically produce field gradients in the three orthogonal directions (dB,/dx, dB,/dy, dB,/dz) over a large (-48 cm) region. By convention, z is the direction of the static magnetic field; the dB,/dx and dB,/dy gradients produce B, $ To whom correspondence should be addressed.
409
0740-3 194/90 $3.00 Copyright 0 1990 by Academic Press, Inc. All rights of reproduction in any form reserved.
410
COMMUNICATIONS
”
0
10
20
0
x (cm)
20
40
60
z (cm)
FIG. I. Profiles of the primary- and cross-field d B / d t generated by one of the transverse gradients (as,/ ax)in the test system. The cross fields are gradients of the fields orthogonal to the primary field and are a required secondary effect of the primary-field gradients. ( A ) Spatial distribution of dB,/dt (the primary gradient), measured at z = 0, as a function of x, the instrument’s horizontal axis. ( B ) Cross-field d B x / d t as a function of z , the instrument’s longitudinal axis, at y = 0. The vertical axis on each curve represents, on a linear scale, the amplitudes used for the human experiments. These graphs show the local time derivative of specific components of the magnetic field as a function of position; the amplitudes used for the human experiments are shown. For the animal study, the values were approximately 10%higher, as indicated in the text.
variations in the plane perpendicular to z . The B, component of these transverse gradients is zero in the center of the patient bore where x = y = 0 and reaches its maximum amplitude at the outermost diameter of the patient bore (Fig. 1A). As a result of the transverse gradients, however, there must arise additional gradients of the transverse magnetic fields, B, and By ( I ). These components (“cross fields”) are relatively constant across the transverse plane but increase with I z J (Fig. 1B). For example, the x gradient must produce, in addition to the desired B, field, a B, field which is roughly constant in x but varies along the z axis. Because the body lies in a position where these fields are at their maximum, the cross fields are more likely than the primary fields to produce electrical stimulation. During the design phase of our instant imaging system, we developed gradient systems capable of creating cross fields varying at up to 76 T/s RMS. Although no medically adverse device effects were noted, we report here that a sinusoidal cross-field d B / d t of 61 T / s RMS produced notable sensations in two volunteers. We believe, on the basis of both the field geometry and overall field magnitude, that the reported sensations were the result of cross-field variations rather than a consequence of the primary d B , / d x gradient. MATERIALS AND METHODS
Gradient System Our gradient coils were designed to have a nominally constant dB,/dx for a 48 cm extent along the magnet z axis. In the present experiments these x gradients were modulated sinusoidally at a frequency of 1.4 kHz for up to 46 ms, repeated at up to 14 pulses/s (70 ms TR). Gradient amplitudes were calculated both from imaging experiments and from the predicted field efficiency and applied current. In the canine studies cross-field
COMMUNICATIONS
41 1
variations of 66 T / s RMS (93 T / s peak) were produced with the x gradient at distances +34 cm from z = 0. Primary gradient field variation, dB,/dt, was 40 T/s RMS at distances of +-20cm from x = 0. For the human studies the gradient system was run at a maximum cross-field amplitude of 6 1 T / s RMS and a primary gradient amplitude of 36 T / s RMS. Additional y gradients with a peak d B / d t 40% of that of the x gradients were also present during this pulsing. The cross fields add in quadrature, however, adding less than 8% to the peak d B / d t , and are ignored in the rest of this report. All experiments were performed within the bore of a 1.5 T imaging magnet. The gradients produced noise of approximately 105 dBA within the patient bore as measured prior in separate experiments.
Animal Handling and Preparation A 35-kg dog was initially anesthetized with intravenous Surital(22 mg/kg) and then intubated with a cuffed endotracheal tube, and maintained on inhalation anesthesia of ha1othane:nitrous oxide:oxygen (0.8- 1%:70%:30%).The animal was ventilated mechanically at 12 breaths/min with a tidal volume of 450 cc. An intravenous of infusion of normal saline was continued to maintain euvolemia. The electrocardiogram was monitored throughout the study using chest and paw leads. It was our intent with the canine study to assess the possibility of adverse reactions resulting from exposure to the time-varying gradients. To maximize the likelihood of these effects we placed the animal so that its heart was at the locations corresponding to peak gradient amplitude. The sinusoidal gradients were pulsed for periods of either 23 or 46 ms; these epochs were repeated at rates of lower than once per minute, once per 2 s, 14 times per second, or at an interpulse delay of 900 ms, corresponding to 80-90% of the animal’s natural cardiac R-R interval.
Human Studies Our initial human imaging experiments were performed on two healthy male volunteers (M.S.C. and R.M.W., who are coauthors of the present Communication). Imaging experiments were performed in either the prone or supine position, on a variety of body parts, thereby leaving the heart, head, or liver at isocenter. During these studies the subjects were kept in constant verbal and visual contact with the experimenters during gradient pulsations. For the human experiments, time-varying magnetic fields, similar to the above, were used and in addition ecg triggered experiments were performed with gradient pulses occurring at both fixed and variable delays for the cardiac R wave. The experiments were repeated with and without earplugs (E-A-R from Carbot Corporation, Indianapolis, IN).
Physiological Monitoring Equipment For the animal studies the ecg was transmitted, via a Hewlett-Packard radio telemetering system, to an Accusynch R-wave detector and finally to a chart recorder. Heart rate was kept at 55 to 72 beats per minute (bpm) for the duration of the experi-
412
COMMUNICATIONS
ment by controlling the anesthetic depth. For one brief period the heart rate climbed to 80 bpm (in the absence of gradient activity). This was corrected by increasing the rate of saline infusion. In the human experiments the ecg was measured through adhesive leads and displayed on a Hewlett-Packard 78352A patient monitor which also was used for gradient triggering.
Gradient Activity Monitor To facilitate data analysis in the animal experiments, an event marker channel of the ecg chart recorder was used to display a trigger signal from the ecg detector indicating the timing of each image acquisition. These event markers, and the presence of gradient pulse-induced artifacts in the ecg record, made it possible to compare accurately the timing of the R wave with that of the gradient pulses. RESULTS
Animal Study In over 500 epochs of exposure to gradient pulses over the entire range of repetition protocols described above there was no evidence in the ecg record of any change correlated with gradient activity. In no case was there visible effect of muscular stimulation or indeed any indication to suggest a physiological response to the magnetic field pulses.
Human Studies Both subjects were surprised to detect a variety of low level sensations during exposure to cross-field gradient pulses of 6 1 T/s RMS. In both individuals the most frequently detected effect of stimulation was described as a small twitch across the bridge of the nose when lying in the prone position with the face pointing directly upward, or in the supine position with the face directed downward. Rotating the head slightly to the left or right reduced or eliminated the stimulation effects. One subject, M., described small twitch contractions at the base of the spine. The second subject, R., experienced similar sensations in the lower back occasionally. R., however, regularly detected twitch contractions on the medial surface of the left thigh. There was no difference in perceived effect for gradients of either 23 or 46 ms duration. The reported effects were small enough that it was not immediately obvious that they were a result of direct stimulation rather than, for example, an acoustic reflex. The results were identical, however, with and without earplugs (which the manufacturer claims to yield more than 40 dB of attenuation at this frequency). Both subjects believed that these sensations resulted from the gradient field exposure. Neither subject reported phosphenes or any other sensory stimulation associated with the exposure to the gradient fields nor was there any evidence of cardiac entrainment or extra systole. Following these reports experimentation at these d B / dt levels was terminated. The stimulation effects, in all cases, were felt simultaneously with the gradient pulses and were transitory, terminating immediately after each gradient pulse. Both
COMMUNICATIONS
413
subjects were subsequently given physical examinations including ecg and cardiac stress tests which indicated no abnormalities attributable to gradient field exposure. When the system was reconfigured for operation at 34 T / s RMS, through suitable modifications of the gradient controller system, the sensory stimulation effects were eliminated completely in these and other volunteers. DISCUSSI 0N
Our initial animal experiments gave no indication for risks to cardiac safety attendant with scanning at the highest d B / d t levels tested. One experimental series, using a TR 90%that of the animal’s R- R interval, was designed to maximize the possibility of generating diastolic contractions or entrainment of the heart to the gradient activity. Neither effect was seen. In these experiments with an anesthetized animal we could not assess the presence or absence of sensory stimulation; only in our human studies was the sensory effect apparent. Though these human data are anecdotal, they do offer some suggestion as to the threshold for direct magnetic stimulation. In prior studies made at 2.0 T with primary gradient field variations of up to 28 T / s RMS and cross fields of up to 18 T/s RMS (note that the gradient geometry of that system was such that the cross fields were of smaller magnitude than the primary fields)no direct stimulation effectswere reported ( 2 , 3 , and unpublished observations) following the collection of over 1 1,000 images on 60 volunteers and patients. Even when that system was operated at 5 1 T / s RMS (primary field) and 32 T/s RMS (cross field) on two volunteers no stimulation effects were observed. However, when that instrument was tested with primary gradient fields up to 57 T / s RMS and cross fields to 35 T/s RMS on two volunteers, one volunteer (who had been noting nasal congestion) described sensations in the nasal sinus area which were not, at that time, attributed to direct magnetic stimulation. At similar d B / d t levels more than 20 canine studies were performed in the setting of segmental myocardial ischemia without any evidence of cardiac entrainment or extra systoles ( 4 ) .Noting the lower threshold for ventricular tachycardia and fibrillation during ischemia, this further supports the safety of the procedure. There exists a considerable body of data on the physiological effects of chronic and acute exposure to time-varying magnetic fields. Recently Reilly ( 5 ) has reviewed the data concerning peripheral stimulation effects resulting from acute exposures. His review data, as well as the results of Bernhardt ( 6 ), predict that at our operating frequency of 1.4 kHz threshold stimulation would occur at peak amplitudes of 90 T / s (64 T/s with sinusoidal gradients). Our present results are in substantial agreement with those predictions. Since our initial submission of this manuscript, Fischer (7) reported that his group saw results similar to our own on their high-speed imaging system. That device differed somewhat, in that the high-speed gradients had their primary field oriented along the z axis; their reported stimulation thresholds of 60 T/ s at 1.25 kHz, however, are close to those reported here. Although the reported and predicted thresholds vary ( 5 ),the levels at which significant safety risks occur, particularly due to direct cardiac stimulation, are likely to be much greater than the sensory or motor thresholds (8).We believe the stimulation effectsreported here to be below
414
COMMUNICATIONS
the level of medical significance though they offer tantalizing clues as to the lower boundaries of direct magnetic stimulation. REFERENCES 1. J. D. JACKSON, “Classical Electrodynamics,”Wiley, New York 1975. 2. R. R. RZEDZIANAND^. L. ~ K E T TRadiology , 161,333 ( 1986). 3. I. L. PYKETTANDR.R. R Z E D Z I A N , MReson. ~ ~ ~ . Med. 5,563 (1987). 4. H. L. KANTOR,R. R. RZEDZIAN, E. BERLINER, P. BEAULIEU, T. J. BRADY,AND 1. L. FYKETT,in “American Heart Association Abstracts,” 1989. 5. J. P. REILLY,Med. Biol. Eng. Comput. 27, 101 ( 1989). 6. J. H. BERNHARDT, Lecture Series 138, Advisory Group for Aerospace Research and Development (NATO), Surseine, France, 1977. 7. H. FISCHER, in “Radiological Society of North America Abstracts 1188,” 1989. 8. D. MCROBBIE AND M. A. FOSTER,Phys. Med. Biol. 30,695 ( 1986).
( 1990)
Sensory Stimulation by Time-Varying Magnetic Fields MARKS . COHEN, *$ ROBERTM. WEISSKOFF,* RICHARDR. RZEDZIAN, * AND HOWARD L. U N T O R ? *AdvancedNMR Systems, Inc., Woburn,Massachusetts 01801;and t Department of Cardiology, Massachusetts General Hospital, Boston, Massachusetts Received September 6, 1989; revised February 14, 1990 When two human volunteers were imaged with magnetic field gradient dB/dt of 61 T / s RMS, the subjects reported, to our surprise, feeling muscular twitches synchronous with gradient pulses over repeated experiments. No adverse or sustained effects were seen. Experiments in a canine, intended to assess the safety of MR imaging with dB/dt of up to 66 T / s RMS, failed to induce detectable changes in the electrocardiogram or to show any signs of gross response to gradient pulsing. Although these data are preliminary, and largely anecdotal, they suggest a level above which such stimulation may occur. We believe that this is the first report of direct human stimulation in an MRI device and that determination of the stimulation threshold may have impact on the selection of appropriate operating points for magnetic imaging systems. o 1990 Academic Press, Inc. INTRODUCTION
In modern whole body magnetic resonance imaging ( MRI) instruments, patients are exposed to a variety of time-varying magnetic fields. While considerable experience has demonstrated the overall safety of these devices, relatively little information is available to indicate the magnitude of the safety margin. In the course of designing our instant imaging equipment, and prior to imaging humans, we performed a series of experiments on a dog, primarily in order to assess the safety of the device. Our initial experiments failed to produce detectable effects and indicated no cause for concern. In our human experiments, however, the two subjects unexpectedly described sensations which we interpret as the effects of direct stimulation. We therefore report here what we believe to be the first description of sensory stimulation by magnetic field gradients of the type used in clinical magnetic resonance imaging systems. While these results are preliminary we feel that their importance in the safe construction of magnetic imaging systems should not be overlooked. To spatially encode the magnetic resonance signal, MR imagers produce timevarying magnetic field gradients which are a small fraction of the main magnetic field. These coil sets typically produce field gradients in the three orthogonal directions (dB,/dx, dB,/dy, dB,/dz) over a large (-48 cm) region. By convention, z is the direction of the static magnetic field; the dB,/dx and dB,/dy gradients produce B, $ To whom correspondence should be addressed.
409
0740-3 194/90 $3.00 Copyright 0 1990 by Academic Press, Inc. All rights of reproduction in any form reserved.
410
COMMUNICATIONS
”
0
10
20
0
x (cm)
20
40
60
z (cm)
FIG. I. Profiles of the primary- and cross-field d B / d t generated by one of the transverse gradients (as,/ ax)in the test system. The cross fields are gradients of the fields orthogonal to the primary field and are a required secondary effect of the primary-field gradients. ( A ) Spatial distribution of dB,/dt (the primary gradient), measured at z = 0, as a function of x, the instrument’s horizontal axis. ( B ) Cross-field d B x / d t as a function of z , the instrument’s longitudinal axis, at y = 0. The vertical axis on each curve represents, on a linear scale, the amplitudes used for the human experiments. These graphs show the local time derivative of specific components of the magnetic field as a function of position; the amplitudes used for the human experiments are shown. For the animal study, the values were approximately 10%higher, as indicated in the text.
variations in the plane perpendicular to z . The B, component of these transverse gradients is zero in the center of the patient bore where x = y = 0 and reaches its maximum amplitude at the outermost diameter of the patient bore (Fig. 1A). As a result of the transverse gradients, however, there must arise additional gradients of the transverse magnetic fields, B, and By ( I ). These components (“cross fields”) are relatively constant across the transverse plane but increase with I z J (Fig. 1B). For example, the x gradient must produce, in addition to the desired B, field, a B, field which is roughly constant in x but varies along the z axis. Because the body lies in a position where these fields are at their maximum, the cross fields are more likely than the primary fields to produce electrical stimulation. During the design phase of our instant imaging system, we developed gradient systems capable of creating cross fields varying at up to 76 T/s RMS. Although no medically adverse device effects were noted, we report here that a sinusoidal cross-field d B / d t of 61 T / s RMS produced notable sensations in two volunteers. We believe, on the basis of both the field geometry and overall field magnitude, that the reported sensations were the result of cross-field variations rather than a consequence of the primary d B , / d x gradient. MATERIALS AND METHODS
Gradient System Our gradient coils were designed to have a nominally constant dB,/dx for a 48 cm extent along the magnet z axis. In the present experiments these x gradients were modulated sinusoidally at a frequency of 1.4 kHz for up to 46 ms, repeated at up to 14 pulses/s (70 ms TR). Gradient amplitudes were calculated both from imaging experiments and from the predicted field efficiency and applied current. In the canine studies cross-field
COMMUNICATIONS
41 1
variations of 66 T / s RMS (93 T / s peak) were produced with the x gradient at distances +34 cm from z = 0. Primary gradient field variation, dB,/dt, was 40 T/s RMS at distances of +-20cm from x = 0. For the human studies the gradient system was run at a maximum cross-field amplitude of 6 1 T / s RMS and a primary gradient amplitude of 36 T / s RMS. Additional y gradients with a peak d B / d t 40% of that of the x gradients were also present during this pulsing. The cross fields add in quadrature, however, adding less than 8% to the peak d B / d t , and are ignored in the rest of this report. All experiments were performed within the bore of a 1.5 T imaging magnet. The gradients produced noise of approximately 105 dBA within the patient bore as measured prior in separate experiments.
Animal Handling and Preparation A 35-kg dog was initially anesthetized with intravenous Surital(22 mg/kg) and then intubated with a cuffed endotracheal tube, and maintained on inhalation anesthesia of ha1othane:nitrous oxide:oxygen (0.8- 1%:70%:30%).The animal was ventilated mechanically at 12 breaths/min with a tidal volume of 450 cc. An intravenous of infusion of normal saline was continued to maintain euvolemia. The electrocardiogram was monitored throughout the study using chest and paw leads. It was our intent with the canine study to assess the possibility of adverse reactions resulting from exposure to the time-varying gradients. To maximize the likelihood of these effects we placed the animal so that its heart was at the locations corresponding to peak gradient amplitude. The sinusoidal gradients were pulsed for periods of either 23 or 46 ms; these epochs were repeated at rates of lower than once per minute, once per 2 s, 14 times per second, or at an interpulse delay of 900 ms, corresponding to 80-90% of the animal’s natural cardiac R-R interval.
Human Studies Our initial human imaging experiments were performed on two healthy male volunteers (M.S.C. and R.M.W., who are coauthors of the present Communication). Imaging experiments were performed in either the prone or supine position, on a variety of body parts, thereby leaving the heart, head, or liver at isocenter. During these studies the subjects were kept in constant verbal and visual contact with the experimenters during gradient pulsations. For the human experiments, time-varying magnetic fields, similar to the above, were used and in addition ecg triggered experiments were performed with gradient pulses occurring at both fixed and variable delays for the cardiac R wave. The experiments were repeated with and without earplugs (E-A-R from Carbot Corporation, Indianapolis, IN).
Physiological Monitoring Equipment For the animal studies the ecg was transmitted, via a Hewlett-Packard radio telemetering system, to an Accusynch R-wave detector and finally to a chart recorder. Heart rate was kept at 55 to 72 beats per minute (bpm) for the duration of the experi-
412
COMMUNICATIONS
ment by controlling the anesthetic depth. For one brief period the heart rate climbed to 80 bpm (in the absence of gradient activity). This was corrected by increasing the rate of saline infusion. In the human experiments the ecg was measured through adhesive leads and displayed on a Hewlett-Packard 78352A patient monitor which also was used for gradient triggering.
Gradient Activity Monitor To facilitate data analysis in the animal experiments, an event marker channel of the ecg chart recorder was used to display a trigger signal from the ecg detector indicating the timing of each image acquisition. These event markers, and the presence of gradient pulse-induced artifacts in the ecg record, made it possible to compare accurately the timing of the R wave with that of the gradient pulses. RESULTS
Animal Study In over 500 epochs of exposure to gradient pulses over the entire range of repetition protocols described above there was no evidence in the ecg record of any change correlated with gradient activity. In no case was there visible effect of muscular stimulation or indeed any indication to suggest a physiological response to the magnetic field pulses.
Human Studies Both subjects were surprised to detect a variety of low level sensations during exposure to cross-field gradient pulses of 6 1 T/s RMS. In both individuals the most frequently detected effect of stimulation was described as a small twitch across the bridge of the nose when lying in the prone position with the face pointing directly upward, or in the supine position with the face directed downward. Rotating the head slightly to the left or right reduced or eliminated the stimulation effects. One subject, M., described small twitch contractions at the base of the spine. The second subject, R., experienced similar sensations in the lower back occasionally. R., however, regularly detected twitch contractions on the medial surface of the left thigh. There was no difference in perceived effect for gradients of either 23 or 46 ms duration. The reported effects were small enough that it was not immediately obvious that they were a result of direct stimulation rather than, for example, an acoustic reflex. The results were identical, however, with and without earplugs (which the manufacturer claims to yield more than 40 dB of attenuation at this frequency). Both subjects believed that these sensations resulted from the gradient field exposure. Neither subject reported phosphenes or any other sensory stimulation associated with the exposure to the gradient fields nor was there any evidence of cardiac entrainment or extra systole. Following these reports experimentation at these d B / dt levels was terminated. The stimulation effects, in all cases, were felt simultaneously with the gradient pulses and were transitory, terminating immediately after each gradient pulse. Both
COMMUNICATIONS
413
subjects were subsequently given physical examinations including ecg and cardiac stress tests which indicated no abnormalities attributable to gradient field exposure. When the system was reconfigured for operation at 34 T / s RMS, through suitable modifications of the gradient controller system, the sensory stimulation effects were eliminated completely in these and other volunteers. DISCUSSI 0N
Our initial animal experiments gave no indication for risks to cardiac safety attendant with scanning at the highest d B / d t levels tested. One experimental series, using a TR 90%that of the animal’s R- R interval, was designed to maximize the possibility of generating diastolic contractions or entrainment of the heart to the gradient activity. Neither effect was seen. In these experiments with an anesthetized animal we could not assess the presence or absence of sensory stimulation; only in our human studies was the sensory effect apparent. Though these human data are anecdotal, they do offer some suggestion as to the threshold for direct magnetic stimulation. In prior studies made at 2.0 T with primary gradient field variations of up to 28 T / s RMS and cross fields of up to 18 T/s RMS (note that the gradient geometry of that system was such that the cross fields were of smaller magnitude than the primary fields)no direct stimulation effectswere reported ( 2 , 3 , and unpublished observations) following the collection of over 1 1,000 images on 60 volunteers and patients. Even when that system was operated at 5 1 T / s RMS (primary field) and 32 T/s RMS (cross field) on two volunteers no stimulation effects were observed. However, when that instrument was tested with primary gradient fields up to 57 T / s RMS and cross fields to 35 T/s RMS on two volunteers, one volunteer (who had been noting nasal congestion) described sensations in the nasal sinus area which were not, at that time, attributed to direct magnetic stimulation. At similar d B / d t levels more than 20 canine studies were performed in the setting of segmental myocardial ischemia without any evidence of cardiac entrainment or extra systoles ( 4 ) .Noting the lower threshold for ventricular tachycardia and fibrillation during ischemia, this further supports the safety of the procedure. There exists a considerable body of data on the physiological effects of chronic and acute exposure to time-varying magnetic fields. Recently Reilly ( 5 ) has reviewed the data concerning peripheral stimulation effects resulting from acute exposures. His review data, as well as the results of Bernhardt ( 6 ), predict that at our operating frequency of 1.4 kHz threshold stimulation would occur at peak amplitudes of 90 T / s (64 T/s with sinusoidal gradients). Our present results are in substantial agreement with those predictions. Since our initial submission of this manuscript, Fischer (7) reported that his group saw results similar to our own on their high-speed imaging system. That device differed somewhat, in that the high-speed gradients had their primary field oriented along the z axis; their reported stimulation thresholds of 60 T/ s at 1.25 kHz, however, are close to those reported here. Although the reported and predicted thresholds vary ( 5 ),the levels at which significant safety risks occur, particularly due to direct cardiac stimulation, are likely to be much greater than the sensory or motor thresholds (8).We believe the stimulation effectsreported here to be below
414
COMMUNICATIONS
the level of medical significance though they offer tantalizing clues as to the lower boundaries of direct magnetic stimulation. REFERENCES 1. J. D. JACKSON, “Classical Electrodynamics,”Wiley, New York 1975. 2. R. R. RZEDZIANAND^. L. ~ K E T TRadiology , 161,333 ( 1986). 3. I. L. PYKETTANDR.R. R Z E D Z I A N , MReson. ~ ~ ~ . Med. 5,563 (1987). 4. H. L. KANTOR,R. R. RZEDZIAN, E. BERLINER, P. BEAULIEU, T. J. BRADY,AND 1. L. FYKETT,in “American Heart Association Abstracts,” 1989. 5. J. P. REILLY,Med. Biol. Eng. Comput. 27, 101 ( 1989). 6. J. H. BERNHARDT, Lecture Series 138, Advisory Group for Aerospace Research and Development (NATO), Surseine, France, 1977. 7. H. FISCHER, in “Radiological Society of North America Abstracts 1188,” 1989. 8. D. MCROBBIE AND M. A. FOSTER,Phys. Med. Biol. 30,695 ( 1986).
