William P. Shuman, MD #{149} David R. Haynor, MD, #{149} George E. Wesbey, MD2 #{149} Daniel J.

Schaefer,

PhD PhD

#{149}

#{149}

Arthur Albert

W. Guy, A Moss,

PhD MD

SuperficialIncreases Exposure

and Deep-Tissue Temperature in Anesthetized Dogs during to High Specific Absorption Rates in a 1.5-T MR Imager

Superficialand deep-tissue heating was measured in five dogs during high-specific-absorption-rate radiofrequency (RF) irradiation to see whether significant temperature changes could be produced by a 1.5T clinical magnetic resonance imager. The RF power output employed was 6.3 times that required for routine imaging. Temperature probes were placed in both deep and superficial tissues, and ternperatures were recorded before, during, and after exposure. In each dog, there was a linear temperature increase of several degrees during RF exposure; the maximal average change was 4.6#{176}C in the urinary bladder. The temperature increase was slightly greater in deep tissues than in superficial tissues. The calculated specific absorption rate, based on the temperature change, averaged 7.9 W/kg for all five dogs. These findings argue for continued caution in the design and operation of imagers capable of high specific absorption rates, particularly when they are used for imaging infants or patients with altered thermoregulatory capability.
Index terms: Magnetic resonance (MR), biologica! effects #{149} Magnetic resonance (MR), experimental #{149} Magnetic resonance (MR), technology Radiology 1988; 167:551-554

1.
projection liver, midpemitoneum, groin, and thigh at the level of each needle tip (arrows) the level of the needle tip demonstrates side.

2. 1, 2. (1) CT-scanned
radiograph of a dog muscle are identified. to confirm its location. its position (arrow). lying on its side. Needles in the CT scans were then obtained (2) CT scan through liver at The dog was lying on its left

Figures

From

the

Department
Hospital,

of Radiology,
1959 N.E.

UniverPacific

sity

of Washington

St., Seattle, WA 98195 (W.P.S., D.R.H., G.E.W., A.A.M.); the Center for Bioengineering, University of Washington, Seattle (A.W.G.); and the Medical Systems Group, Genera! Electric Company, Milwaukee (D.J.S.). Received October 2, 1987; revision requested November 6; revision received November 30; accepted December 22. Supported in part by a grant from Genera! Electric Medical Systems. Address reprint requests to W.P.S. 2 Current address: Department of Radiology, Scripps Memorial Hospital, La Jolla, Calif.
C RSNA,

1988

(RF) radiation can cause heating in living tissue. The current Food and Drug Administration (FDA) recommendation states that RF radiation exposure duning magnetic resonance (MR) imaging of humans represents an insignificant risk if it does not exceed a whole-body average specific absomption mate of 0.4 W/kg (1). This rate was derived from reports indicating that the onset of thermal effects (work stoppage in laboratory animals) took place at 4 W/kg; to derive a broad safety margin, this value was divided by a factor of 10 (1). More mecent reports have suggested that, under certain conditions, a higher specific absorption rate (0.4-4.0 W/kg) may be used for MR imaging in humans without resulting in dangerous tissue heating (2-6). In these studies temperature changes in animals and humans were measured immediately before and after (but not during) imaging. Temperatures were measured in the mouth (sublingual pocket) and esophagus and on the surface of the skin; oral and esophageal temperatures increased by up to 0.5#{176}C,and surface skin temperatures by up to 3.5#{176}C.These findings are consistent
ADIO-FREQUENCY

with theory, which predicts greater RF-induced heating superficially than in more internal locations (2, 6-9). We undertook this study to measure both superficialand deep-tissue heating during imaging at a specific absorption rate that was much higher than is normally used clinically but that was still achievable with a clinical MR imager. To accomplish this, it was necessary to use an animal that is smaller than adult humans and to bypass both the RF power monitor of the clinical MR imager and the existing software, which has built-in safeguards high designed specific to prevent absorption rates. the use of

MATERIALS
We studied
weighing 30 mg/kg venously esthesia; ministered ment was

AND
five healthy

METHODS
mongrel
was level the

dogs
given intraof anwas adexperi-

18-28 kg. Each dog of sodium pentobarbital to induce a surgical an additional to each and dog 15 mg/kg during

to maintain intubated

anesthesia. allowed

Each dog to breathe was moniintervals. 16-gauge for placing

room air. The ventilatory mate tored at approximate 2-minute With each dog, long (15-cm) needles, originally designed

551

therapeutic tors spinal (Deseret neck percutaneously

hyperthermia Medical; positioned muscle and region), axilla and/or midperitoneum and the and Sandy,

sensor Utah), in the

therntiswere paramuscle,

thigh

the midliver mesentery tissues

(gut subcutaneous (midline).

of the

groin

After the needles were placed, the position of each needle tip was confirmed with computed tomography (CT). A CTscanned projection radiograph (ScoutView; General Electric, Milwaukee) was first obtained to locate each needle tip (Fig. 1), and then axial CT scans were obtamed at the level the of each position tip (Figs. or 2-4). more Occasionally, of one

needles
tion Anesthesia placement

was
confirmed

adjusted
by

and
means

the

new

posi3. 4.

of CT. just before so that it contin-

was induced of the needles

Figures

3, 4.

ued
imaging

for the
suite

duration

of the

CT study

(18-

midperitoneum neous tissue

(3) CT scan through the dogs abdomen (arrow). (4) CT scan demonstrates the of the groin (arrow).

at the position

level of the needle tip of the needle in the

in the subcuta-

25 minutes),

the move to the adjacent MR (3 minutes), and the setup

of the

MR imaging

experiment
time, no regulate

(12-15
attempt or mainover as the lonout tisa

ment,

the

temperature

recording

system

minutes). was made tam the Each

During this to externally

was

located

just

outside

the

magnetic

tions; eight mm intervals;

5-mm-thick and a 256

sections X 256

with matrix. pulse

2.5-

dogs body temperature. 16-gauge needle fit snugly catheter twice as long metal needle was scored

shield of the MR imager fore in a magnetic field even though it was just center We operating coil. this use

and was thereof only 0.003 T 5.5 m from the magnet. 1.5-T Electric)

Temperature
15 minutes

recording
after the end

was

continued
se-

for

of the

blind needle;

plastic the

of the bore of the 1.5-T used a standard clinical MR imager with (General at 64 MHz

quence. In dog
absorption

3 we also
mate by

estimated
computing

the
the

specific
net

gitudinally so that it could be of tissue (leaving the catheter sue), split longitudinally, and from the plastic ing the catheter thermistor could catheter be slid after sensor

backed in the removed

whole-body

a quadrature

peak absorbed sequence, the

load-diverted

power. forward,
RF power

During the pulse reflected, and

was measured.

catheter without tip. A high-resistance on the end

dis!odg-

The software system had of a calculated

and power monitor of been modified to allow specific absorption mate A research was employed, to institutions version a that

The
the

net
same

peak
RF

coil
magnetic

loss

power
field

(measured
strength and

at

of a wire blind

down to the end of the the needle was removed;

greater than 0.4 W/kg. of operating software version available only have approval subjects review to greater than

determined tom) was levels with

with subtracted

a small, after

lossless phanthe peak power levels Since

depth marks on the catheter and the thermistor wire were used to ensure positioning of the thermistor at the end of the catheter. In three inserted a Foley distended ture saline the bladder of the into the catheter, dogs, a thermistor was through was in

from their local board to expose 0.4 W/kg. This

human humans software monikey-

were converted use of the pulse

into average waveforms.

the

TR was

known,

dividing

the

net

peak
dog

allows the specific-absorption-rate tors to be overridden with

absorbed (22 kg)

power yielded

by the mass of the the specific absorption

a specific

rate

in watts

per

kilogram.

urinary bladder after the bladder

stroke

command.
ambient temperature at 18#{176}-20#{176}C with was a relative bore to

with 50 mL of room-temperasolution. The saline was left for approximately 8 minutes

The room maintained humidity fan was provide on

RESULTS
The respiratory rate was somewhat depressed (to four to eight breaths per minute) in all of the dogs during the early portion of the experiment, presumably because of the anesthesia. In two dogs, the respiratory rate accelerated to 60 breaths per minute toward the end of the pulse sequence. During the initial 15-minute temperature monitoring period (no RF), all probes recorded stable temperatumes in all dogs (change was <0.3#{176}C) (Fig. 5). During the pulse sequence (RF on), there was a linear temperature increase for all probes in all dogs (Fig. 5). The range of ternperature increase was 2.4#{176}-4.7#{176}C (average, 3.7#{176}C)for neck muscle, 1.3 #{176}4.3#{176}C (average, 2.5#{176}C)for thigh musdc, 3.2#{176}-5.6#{176}C (average, 4.1#{176}C)for the liver, 3.2#{176}-5.3#{176}C (average, 4.4#{176}C) for the midperitoneum, 2.i#{176}-5.3#{176}C (average, 3.8#{176}C)for subcutaneous axillary tissue, 2.1#{176}-5.3#{176}C (average, 3.4#{176}C) for subcutaneous groin tissue, and 3.4#{176}-5.6#{176}C (average, 4.6#{176}C)for

of 35%-40%; the magnet during each experiment

before

the

start

of scanning

in order
with

to
body

achieve partial equilibration temperature. Although the high-resistance tors ture were designed in a high-RF to measure environment

mm.

After

a measured airflow the high-resistance

of 0.10 m3/ thermistor

thermistempera(30-1, of tissue 000

sensors were positioned, tized dog was placed with its body within

each anesthein the magnet bore the confines of the

RF coil.
the cm.) dogs

(The
varied

coil
in

was

150 cm long,
from 80

and
to 100

MHz),

placing

them

in snug-fitting
some degree from adjacent

plas-

length

tic catheters electrical

provided insulation

and minimized any induced current in the thermistor wires. In addition, the terminal 30.5 cm of the connection was high-resistance carbon-impregnated Teflon rather than wire. The thermistor wires were connected by a 5.5-rn cable to a BSD-300 mobile computerized temperature recording system designed to collect, display, and record tissue temperature data (BSD during Medical therapeutic Corporation, hyperthermia Salt Lake thermistor before
calibration

Baseline temperatures were recorded every minute for 15 minutes from each of the six thermistor sensors (Fig. 5). A 25.6minute maxima! (16-kW) RF power output pulse sequence was then performed (with only 0.2 dB of RF transmit attenuation). minute Temperatures were

recorded

every

pulse cause timal

during sequence of an imaging

the sequence. lasted 35.6 imager for malfunction. these dogs

In dog 5, the minutes beSince opwould have

City). The high-resistance were computer calibrated

periment with a probe

used an RF transmit attenuation of about 8.0 dB (for a 90#{176} pulse), the power output for this maximal sequence (no transmit attenuation) was about 6.3 times greater than that required for routine imaging. The spin-echo pulse sequence employed repetition time (TR) of 1,000 msec; echo times of 20, 40, 60, and 80 msec; six repeti-

probes each exbath by the probe experi-

and
National

a standard
Bureau

probe calibrated of Standards. The 0.1#{176}C. During each

accuracy

was

552

Radiology

May

1988

.,

N*p:i;o

TEMP

:
,.

Table 1 Specific Absorption

Rate

Based

on Average

Temperature
Average Temperature Change (#{176}C) 2.6
2.9 3.6

Change
Specific Rate Absorption (W/kg)* 5.9
6.5 8.1

\ 0>7 .
.0.

Dog 1
2 3

Weight 28
26 22

(kg)

Duration of RF Pulse Sequence (mm) 25.6

25.6 25.6

.55:55:55.P

TIME

(MINUTES)

4
5
*

18
22

25.6
35.6

4.7
5.1

10.5
8.3

Figure 5. over time

are plotted The temporal relationships of the temperature increase

Temperature (TEMP) changes in dog 2. For clarity, temperatures for only four of the six tissues.

The average

rate was 7.9 W/kg.

and the linearity during RF expoCooling mechanisms in dogs are quite different from those in humans. Dogs lose heat mostly by panting, not by sweating (12). In addition, dogs have a heavy coat of hair as external insulation and, unlike humans, cannot remove insulation to increase heat loss. Our dogs initially experienced ventilatory suppression, probably because of the anesthetic agent used (pentobarbital), so they had little capacity to get rid of heat from RF irradiation. (Pentobambital by itself can increase the come temperature in dogs when given as a bolus, but at a much slower mate, by tenths of a degree per hour [13].) Thus, our dogs constituted a near worst-case situation from the standpoint of heat loss. In addition, these anesthetized dogs had little capacity to generate heat or maintain body temperature. They were also small enough to fit within the confines of the RF coil. Thus, their entire bodies were irradiated and heated. This situation might simulate that of a sedated small human patient (infant) who is wrapped in blankets (unable to evaporate sweat and well insulated). The high-specific-absorption-rate exposures for these experiments were also designed to be a near worst-case situation from the standpoint of heat deposition. The high level of RF power used was much greater than that required for typical imaging in these dogs or in many smallto average-sized humans. Imaging in heaviem humans may require maximal RF amplifier power, similar to what we used in this experiment, but in such individuals the specific absorption rate would typically be lower because of the greater body mass. In our dogs, the lowest rate was in the largest dog (Table 1). Commercial clinical MR imagers have safeguards to prevent the madvertent use of excessively high RF power. One such safeguard is the software-mandated calculation of the specific absorption rate for each patient based on the sequence selected and the exceeding specific patients weight. Sequences the FDA guidelines for absorption rate are rejected.

sure seen in this graph were typical. Note the relatively small change in the periods

before

and

after

RF exposure.

the urinary bladder. The estimated mean specific absorption rate (assuming no heat loss), which was derived by averaging the temperature changes for all probes in each dog (using an average tissue-specific heat of 0.83 kcal [3.47 kJ]/kg/ #{176}C), ranged from 5.9 to 10.5 W/kg (Table 1). For dog 3, the calculated specific absorption mate based on the average tempematume change for all tissues was 8.1 W/kg, and the mate determined by measuring the absorbed power in the dog-coil combination was 7.1 W/kg.

DISCUSSION A reduced ability to perform tasks

has been observed in rats and monkeys whose body temperatures were increased by 1 #{176}C after whole-body RF exposure of 4 W/kg (10). In humans, a prudent specific-absorptionrate threshold of 1.5 W/kg for longduration studies (20 minutes or more) and 4 W/kg for studies of 10 minutes or less has been suggested (2). This suggestion is based on a basal metabolic mate in humans of 1.5 WI kg during sleep and 15 W/kg during heavy exercise. Other studies have suggested that 4 W/kg may be themmally acceptable for longer periods (3, ii). With high metabolic rates, such as those that occur during exemcisc, healthy humans are able to dissipate heat through the evaporation of sweat over a large skin surface area and, to a lessen extent, through an increased respiratory rate, so that the come body temperature does not rise significantly. The cooling of skin due to evaporation of sweat is dependent on ambient temperature, humidity, and airflow. Controlling core ternpemature with these mechanisms is important, since a rise of 4#{176}C or greater can produce heat stress, cellulam damage, and even death. 2

The peak RF power needed to image a patient is also calculated (based on mass and anatomic landmarks) and is downloaded to an RF power monitor; this monitor shuts down the system if the peak RF power, the pulse widths, or the duty cycle are excessive. Another safeguard is present in the technique used to increase RF power output; the output can be increased only in small steps (as limited by the software), so that a single keystroke error cannot result in unintended high exposure. Skin is a particularly good organ for dissipation of heat, since skin blood flow can be increased by a factom of 20 in response to a hyperthermic environment (14). Such increased skin blood flow is dependent on an intact and physiologically normal cardiovascular system. Skin cooling that results from increased blood flow is augmented by the evaporation of sweat. Good skin and subcutaneous tissue cooling is important, since theory predicts that a propomtionally greater RF heat deposition from MR imaging will occur near the surface in cylindrical structures (2, 49). On the basis of this theory, investigators have recently suggested that it is more important to assess surface temperature than deep-tissue tempemature when evaluating the heating induced by RF irradiation during MR imaging (6). Deep-tissue heat is redistributed and carried away by an intact cardiovascular system as well. However, heat (such as that generated subcutaneously) can also be carried to deeper tissues by the cardiovascular system; the degree to which heat generated
superficially is carried to deep tissues

(thermodynamic stirring) is unknown. Subcutaneous fat is a good electrical insulator that may somewhat isolate deeper internal current loops generated by RF irradiation (15). This electrical insulating effect

Volume

167

Number

Radiology

#{149} 553

of fat may alter the proportional position of RF-induced heating

tween superficial and deeper

debetissues

of deeper than that

from that predicted by theory or measured with a single-tissue phantorn. Subcutaneous fat is also a good thermal insulator, so that heat deposited in or carried to deeper body tissues is lost more slowly than heat from superficial tissues. Some regions of the body are not
perfused by blood, ed in these regions and may heat depositdissipate

tissues was slightly greater of subcutaneous tissues. Fourth, the heating of urine in the bladder was similar to the heating of other deep tissues (Le., the liver and
rnidperitoneum).

2.

Budinger TF. Nuclear (NMR) in vivo studies: health effects. J Comput

5:800-811.

magnetic known Assist

resonance thresholds
Tomogr

for
1981;

3.

These data suggest that thermal diffusion and blood flow may make the dog body a stirred thermodynamic system, in which gradients in
RF power deposition may not temperaresult

Schaefer DJ, Barber BJ, Gordon CJ, Zielonka J, HeckerJ. Thermal effects of magnetic mewnance imaging (MRI)(abstr.). In:Book of abstracts: Society of Magnetic Resonance in Medicine 1985. Vol. 2. Berkeley, Calif.: Sodeof Magnetic Resonance in Medicine, 1985;
FG, Crues JV. ous blood flow related ing magnetic resonance (abstr.). Clin Res 1986; Shellock FG, Schaefer responses to different Shellock

4.

5.

in equally
superficial

large
and

gradients
internal

between
6.

more slowly. These regions include the lens of the eye, urine in the unnary bladder, and ascitic fluid within the peritoneurn. The temperature threshold for generating cataracts in the lens of a rabbits eye is around 42#{176}C (16); however, a local specific absorption rate of about 100 W/kg is required to generate cataracts (17). It is unknown whether local hot spots might develop during high-specificabsorption-rate irradiation in unperfused fluid collections.

tunes. The data also indicate that without existing safeguards, clinical MR irnagers could produce significant heating in some tissues under some conditions. Thus, continued caution in the design and use of these imagers is probably warranted,
particularly in units modified so they

7.

8. 9.

can be operated rates above the lines.

Extrapolating

at specific absorption current FDA guidethese data from anes-

10. 11.

Several observations from our experiments.

high-specific-absorption-rate

can be made First, this

se-

quence produced by a clinical MR irnager increased the temperature in all of the tissues measured by several degrees; the greatest average increase was 4.6#{176}C in the urinary bladder. The maximal temperatures were 42.3#{176}C in subcutaneous tissue (groin, dog 5), 41.6#{176}C in muscle (neck, dog 3), 43.4#{176}C in the liver (dog 5), 43.2#{176}C in the midperitoneum (dog 5), and 43.4#{176}C in the urinary bladder (dog 5). In all dogs, at least one temperature measurement was above 41.5#{176}C. Second, the temperature increase during
RF irradiation was relatively linear

thetized dogs to unanesthetized humans would not be valid. However, we believe that particular attention should be paid to specific absorption rates in the imaging of infants; patients with heart conditions, altered thermoregulation due to fever or
drugs, or anhidrosis; and patients

Alterations in cutaneto tissue heating durimaging at 1.5 Teals 34:344A. DJ, Crues JV. Thermal levels of radiofrequency power deposition during clinical magnetic resonance imaging at 1.5 Teals (abstr.). Magn Reson Imaging 1986; 4:94. Shellock FG, CruesJV. Temperature, heart rate, and blood pressure changes associated with clinical MR imaging at 1.5 T. Radiology 1987; 163:239-262. Bottomley PA, Andrew ER. RF magnetic field penetration, pulse shift, and power dissipation in biological tissue: implications for NMR imaging. Phys Med Biol 1978; 23:630643. Bottom!ey PA, Edelstein WA. Power deposition in whole-body NMR imaging. Med Phys 1981; 4:510-512. Bottomley PA, Rowland WR, Edelstein WA, Schenck JF. Estimating radiofreuency powem deposition in body NMR imaging. Magn Reson Med 1985; 2:336-349. Michaelson SM. Microwave biological effacts: an overview. Proc IEEE 1980; 68:40-49. Adair ER, Berglund LG. On the thermoregulatory consequences of NMR imaging. Magn Reson Imaging 1986; 4:321-333.
Sugano Y. Seasonal changes in heat balance

12. 13.

14. 15.

who cannot communicate well enough to describe sensations of warmth. If specific absorption rates exceed the FDA recommendations, care should be taken not to wrap patients tightly with blankets or restraints. In addition, relatively low ambient room temperature and humidity should be maintained, along with good airflow through the rnagnetbore. U
References
1. Radiological Health Bureau of US.A. Guidelines for evaluating electromagnetic exposure for trials of clinical NMR systems. Washington, D.C.: Food and Drug Administration, 1982.

16. 17.

of dogs acclimatized to outdoor climate. Jpn J Physiol 1981; 31:465-475. Baum D, HalterJB, Taborsky GJ, PorteD. Pentobarbital effects on plasma catecholamines: temperature, heart rate, and blood pressure. Am J Physiol 1985; 248:E95-E100. Rowell LB. Cardiovascular aspects of human thermoregulation. Circ Rca 1983; 52:367-379. Guy AW, Webb MD, Sorenson CC. Determination of power absorption in man exposed to high frequency electromagnetic fields by thermographic measurements on scale modcia. IEEE Trans Biomed En 1976; 23:361-371. Adey WR. Tissue interactions with nonionizing electromagnetic fields. Physiol Rev 1981; 61:435-514. Guy AW, Lin JC, Kramer P0, Emery AF. Effect of 2,450-MHz radiation on the rabbit eye. IEEE Trans Microwave Theory Techn 1975; 23:492-498.

over

time

(Fig.

5). Third,

the

heating

554

#{149}

Radiology

May

1988