This Article Abstract Figures Only Full Text (PDF) Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Condon, B Articles by Hodgson, R Search for Related Content PubMed PubMed Citation Articles by Condon, B Articles by Hodgson, R

The ferromagnetic pillow: a potential MR hazard not detectable by a hand-held magnet

Departments of ¹Clinical Physics and ²Neuroradiology, Institute of Neurological Sciences, 1345 Govan Road, Glasgow G51 4TF, UK

	Abstract

Top Abstract Introduction Methods and materials Results Discussion Conclusions Appendix References

 
This paper describes an incident in which an apparently normal hospital pillow became a ferromagnetic missile when brought into the proximity of a 1.5 T MR system owing to a fine internal spring system within the pillow. Measurements revealed that the 1 kg pillow reached a maximum velocity of 33.7 km h^-1 after undergoing a maximum acceleration of 9.9g. Non-pathological cervical spines should sustain the measured forces and torques without significant injury. However, the effect could be injurious or even fatal to patients suffering from an existing cervical instability, for example due to rheumatoid arthritis. Of more general concern is the fact that the use of a powerful hand-held magnet did not reveal the presence of ferromagnetic components in this instance. Large objects containing sparsely distributed ferromagnetic materials may not be deflected by such a magnet but could still represent a hazard in the MR environment.

	Introduction

Top Abstract Introduction Methods and materials Results Discussion Conclusions Appendix References

 
The so-called "ferromagnetic missile effect" occurs when an object containing ferromagnetic material is brought into the proximity of an MR system. If the object contains sufficient ferromagnetic material and is unrestrained, it will undergo rapid acceleration and may cause injury if it strikes patients or staff. Objects responsible for injury cover a wide range, including, for example, a hairpin [1] to the 50 kg extension tines of a forklift truck [2].

Objects being taken in to the high field environment that are suspected of containing metal can be tested for ferromagnetism by bringing them into proximity with the MR system itself, provided that no patients or other staff are in the magnet room. If forces of attraction become evident, the object is regarded as a potential missile. However, testing objects in this way can itself pose an element of risk because forces and torques can vary very rapidly over small distances in the vicinity of shielded magnets [3]. Strains to fingers, hand and wrist can result from their sudden onset. Changes in such forces as well as damage to the magnet can occur if objects are torn from the operator's grasp. In response to this, it is common practice to use hand-held magnets with fields of up to several hundred millitesla to test objects in a more controlled manner outside the magnet room.

This report describes an incident in which a non-ambulatory patient was brought to the MR unit for investigation of thoracic cord compression. The patient had previously been transferred from their hospital bed to the non-ferromagnetic MR patient trolley. The patient was in considerable pain, so two pillows from the patient's bed were added to the pillow already on the MR trolley to make the patient more comfortable for the transfer. On arrival at the MR unit the patient was taken through the usual MR safety checklist and no contraindications were revealed. The patient was to be introduced into the magnet head first, but as she was being transferred to the examination table one of the pillows (the middle of the three) rotated and rapidly accelerated into the magnet. After passing through the isocentre it decelerated until it reversed its direction of motion. It then oscillated back and forth along the bore of the magnet until frictional forces brought it to rest close to the isocentre.

The patient was not injured and in fact was unaware that anything had happened, as the loss of pillow height occurred as she was being transferred onto the table.

The pillow was forcibly removed and the examination continued without incident. The pillow was taken out of the magnet room where it was tested with a hand-held horseshoe magnet producing a field of 237 mT at the pole pieces. No force of attraction was noted even when the magnet was pressed deeply into the pillow. Suspending the pillow from a string and bringing the magnet into proximity also failed to demonstrate any deflection. The pillow was carefully palpated yet revealed no hard objects. The pillowcase in which it was enclosed was sewn and so had no zip, metal or otherwise. The pillow was indistinguishable in sight and touch from a standard hospital pillow. It weighed 0.998 kg, compared with the usual type of hospital pillow, which has an average weight of 1.011 kg.

The pillow was cut open for further inspection. As shown in Figure 1, the pillow contained a pack of 40 fine springs. The pack was surrounded on all sides by padding material that, when not confined, had a minimum thickness of 10 cm. It was presumably this padding, as well as the small diameter of the wire (1 mm), that prevented the springs being detected by palpation. Each spring consisted of approximately 5.5 turns, 46 mm in diameter, giving a total wire length of 800 mm and a mass of 4.1 g. The total mass of metal in the pillow was 164 g, amounting to 16% of the mass of the pillow.

View larger version (126K): [in this window] [in a new window]   Figure 1. Photograph showing the pillow, the pack of 40 springs found within the pillow and an example of a single spring taken from the pack.

Further measurements were performed to ascertain the forces on the pillow and to determine whether these could possibly represent a hazard. In particular, the concern was addressed that a patient with the pillow under their head being introduced feet first in to the magnet may experience a hyperflexion injury.

	Methods and materials

Top Abstract Introduction Methods and materials Results Discussion Conclusions Appendix References

 
Forces experienced by ferromagnetic objects depend on many factors, including field strength as well as the design, type and effectiveness of the static field shielding. The following measurements and calculations should therefore be regarded only as illustrative examples of possible forces. The particular system to which these measurements apply is a passively shielded 1.5 T Siemens 63SP Magnetom (Siemens, Erlangen, Germany). Measurements were made from two positions on the couch top. The first position was from the notional "creep point". This is the point furthest from the magnet at which the pillow would accelerate within 30 s of being placed there. Objects moving from this point will reach a maximum velocity and therefore produce maximum im2pact. The second position was the point at which the force of attraction on the pillow was greater than the restraining force of a human head resting on it, called the flexion point. This was established by putting a 90 kg subject head first (to avoid a hyperflexion injury) into the magnet. The point at which the force on the pillow overcame the weight of the head and the frictional forces was recorded.

To estimate velocities and accelerations, a calibrated medical strobe light operating at its maximum rate of 60 flashes per s, was used in conjunction with manual exposure of photographic film. Placing this apparatus within the bore of the magnet would not have been practical, as the measurement apparatus itself contained ferromagnetic material and in any event was likely to be damaged by the rapidly moving pillow. The velocity and acceleration of the pillow were therefore determined indirectly by attaching a white ball to the pillow via a non-extensible terylene thread. The mass of the ball and thread was only 3.4 g, compared with the 0.998 kg mass of the pillow, and therefore was not felt to significantly alter the dynamics of the experiment. The ball was allowed to travel freely against the background of a calibrated scale.

The resulting exposures showed images of the ball at each flash, which allowed the velocity of the ball to be calculated. Acceleration was estimated from the separation of three successive flashes of the moving ball, assuming constant acceleration.

To compare the results with the literature pertaining to the biomechanics of spinal injuries, estimates were also made of other commonly used parameters, namely force, torque and angular acceleration (see Appendix for the basis of calculations).

Finally, the potential maximum impact force from the pillow to a human head already in the magnet was calculated. For this, the calculated maximum speed (v_max) that the pillow reached from the creep point was used and was simply based on the maximum distance the ball travelled between successive flashes. The assumption is made that the full kinetic energy of the pillow (calculated as mv_max²/2, where m is the mass of the pillow) is transferred to a head that is in the isocentre of the magnet. At this point the pillow will be experiencing no forces of attraction in the highly uniform field and so will be travelling at a constant velocity.

	Results

Top Abstract Introduction Methods and materials Results Discussion Conclusions Appendix References

 
The notional creep point was found to be 0.73 m from the entrance to the magnet bore. The point at which the forces on the pillow overcame those of a resting head plus friction, the flexion point, was 0.1 m inside the magnet as measured from the bore entrance. The entrance to the bore is 1.38 m from the magnet isocentre on this larger style of passively shielded 1.5 T magnet.

Inspection of the images revealed that the pillow underwent acceleration up to approximately 0.4 m inside the magnet, at which point it maintained a constant velocity within measurement error. This reflects the fact that translational force is related to the difference in field experienced by an object. In the homogeneous field in the centre of the magnet there will be no translational force and thus no acceleration. The maximum velocity from the flexion point expressed, as the mean and standard deviation of three repeat measurements, was 7.38±0.26 m s^-1 (26.6±0.9 km h^-1).

The maximum velocity attained from starting at the creep point was 9.4 m s^-1 (33.7 km h^-1). In comparison, a pair of 23 g disposable metal scissors reached a maximum velocity from the creep point of 14.3 m s^-1 (51.5 km h^-1), and a 91 g metal spectacles case reached a maximum velocity of 21.6 m s^-1 (77.8 km h^-1).

In the first 33 ms interval, the pillow underwent an acceleration of 57.6 m s^-2, approximately 5.9 times the acceleration g due to gravity. Over thenext interval, acceleration increased to 61.2 m s^-2, which suggested that the assumption that acceleration over the first set of three flashes was constant was a reasonable approximation. During the next 33 ms interval, however, acceleration increased to 97.2 m s^-2 (equivalent to 9.9 g). The main field does not fall off in a non-linear manner along the axis of the magnet and this is additionally confounded by the shielding distribution. The force on a ferromagnetic object will therefore vary non-linearly along the axis of the bore.

The initial force on the pillow was 57.5 N at the flexion point. Torque to the head at the flexion point was estimated to be 8.6 N m, producing an angular acceleration of 85.2 rad s^-2 in a typical adult head of 4.5 kg. The maximum impact pulse of the pillow accelerating from the creep point to strike a patient's head positioned in the centre of the magnet was estimated to be 44.1 J.

	Discussion

Top Abstract Introduction Methods and materials Results Discussion Conclusions Appendix References

 
In general, if a head is subject to a sufficiently large force that is forwards and upwards from a blunt impact to the occiput from below (comparable with the case where a patient with the head resting on a ferromagnetic pillow is introduced feet first in to the magnet), a disruptive hyperflexion injury may potentially result. In such an injury, spinous processes can be distracted, with posterior ligaments receiving the greatest trauma. If a more severe force has been applied, posterior ligaments can be torn and articular surfaces of the apophyseal joints may lose all contact and may override. If trauma occurs in hyperflexion, locking of the articular surface processes can occur bilaterally [4]. Sudden flexion may damage the spinal cord either by immediate mechanical stresses or by compression due to herniation and protrusion of one or more nucleus pulposus [5].

We will now attempt to establish whether the forces and torques calculated for the pillow may be significant for non-pathological human spines. Comparison is made with the extensive literature of the biomechanics of spinal cord injury.

In a study of intact human cadavers, flexion failures occurred with forces of 667 N [6]. In isolated functional segments of C1 and C2, rupture of the transverse ligament occurred at 117–1765 N (mean of 20 specimens, 824 N) [7]. Forces required to fracture the odontoid ranged from 687–1765 N. In comparison, the initial force at the flexion point for the pillow was only 57.5 N (49% of the lowest level required to produce injury) and is therefore unlikely to be significant in the normal spine.

It has been estimated that there will be a 50% chance of concussion to a 1300 g brain if it experiences an angular velocity of 50 rad s^-1 or angular acceleration of 1800 rad s^-2 [8]. Again, this is well above the 85 rad s^-2 acceleration calculated for the pillow.

Flexion-only (i.e. no lateral component) cadaver studies have applied torques of up to 189.9 N m without significant damage being recorded [9], although when there is some lateral component, injuries may occur with torques as low as 22.6 N m [10]. Some lateral torque is possible, as rather than pushing the head forward the pillow may, if placed asymmetrically under the head, push the head both forward and to one side. None the less, the torques recorded with the pillow were only 8.6 N m (38% of the level required to produce injury) and so are unlikely to produce damage in the normal spine.

Superoinferior vertex impacts on unembalmed corpses produced vertebral fractures with peak forces of 5700 N (corresponding to an initial impulse of 380 J) for normal structures but 3600 N (corresponding to an initial impulse of 250 J) in cadavers with abnormal structures [11]. Even in abnormal spines, a blow by the pillow to the vertex of 44 J (18% of the level required to produce injury) will not be significant.

Whilst it would appear that the ferromagnetic pillow is unlikely to present a significant hyperflexion danger to the normal neck, the threat to patients with existing instability could be far more serious. In the words of Patterson [12], "the person with an unstable atlanto-axial joint is leading a rather precarious life, sometimes with fatal outcome. An injury superimposed on this condition may cause serious disability and death". Deaths associated with spontaneous events or trivial movements have been reported from a number of sources [12–18]. Causes of such instability include fracture and the congenital absence or hypoplasia of the dens. Rheumatoid arthritis [19] is the most common cause. Patients in whom the condition is diagnosed may be given neck braces to help maintain stability. However, these are routinely removed before MR, particularly if they contain metal.

It is also worth noting that the moving pillow certainly has enough momentum to pull out any injection, infusion or invasive blood pressure lines associated with the patient.

It is also important to note that the forces and torques estimated here may represent close to the minimum forces at 1.5 T because most modern systems are actively shielded, with a generally more rapid decrease in static field strength along the axis (called the static field gradient [3]). Translational forces on ferromagnetic objects in such actively shielded systems will therefore be higher than measured on this older style, passively shielded system. Similarly, ultrahigh (>2 T) field systems, where the shielding is increased to bring fringe fields down to volumes similar to 1.5 T systems, will also have higher static field gradients and will be responsible for greater forces on ferromagnetic objects.

That a powerful hand-held magnet did not reveal the presence of ferromagnetic material in the pillow raises wider concerns. MR systems produce high magnetic fields over considerable volumes and all parts of even large objects such as the pillow will experience the field. Hand-held magnets on the other hand may produce fairly high fields in proximity to the pole pieces, but these fall away rapidly. For example, in the case of the 237 mT hand-held magnet used here, thefield falls to less than 10 mT at a distance of only 5 cm. Thus, even when pressed into the pillow, only a small fraction of spring material will experience a strong static field gradient, thus no measurable deflection of a 0.998 kg pillow is produced.

Hand-held magnets should therefore not be considered a foolproof means of identifying ferromagnetic missile potential in large objects, as they are unlikely to reveal the presence of significant amounts of metal if this is sparsely distributed within the object. Such objects, which include furniture as well as bedding materials, instead need to be assessed in proximity to the MR system, even though this can itself involve some element of hazard and should certainly not be attempted if any patients or other staff are in the magnet room.

	Conclusions

Top Abstract Introduction Methods and materials Results Discussion Conclusions Appendix References

 
Some pillows contain sufficient ferromagnetic material to make them an MR hazard to patients with existing cervical spine instability.

Hand-held magnets may not be able to demonstrate the potential of large objects as ferromagnetic missiles if metal is distributed sparsely within the objects.

	Appendix

Top Abstract Introduction Methods and materials Results Discussion Conclusions Appendix References

 
Force was calculated as the product of the mass of the pillow and its acceleration, and torque as the product of force and lever arm length. Lever arm length was taken as the distance from the occiput to the angulation point of the neck (approximately 0.15 m).

Angular momentum L was calculated from:

where w is the angular velocity, r is the lever arm length and m is the mass.

However, as torque T is also related to the rate of change of angular momentum as:

This allows dw/dt (angular acceleration) to be calculated.

Received for publication January 18, 2001. Revision received April 9, 2001. Accepted for publication May 14, 2001.

	References

Top Abstract Introduction Methods and materials Results Discussion Conclusions Appendix References

 

1. Villegas E, Valldeoriola F, Trilla A. Untoward effects of MR imaging. AJR 1994;162:997.
2. Fowler J, ter Penning B, Syverud SA, Levy RC. Magnetic field hazard. N Engl J Med 1986;314:1517.[Medline]
3. Kanal E, Shellock FG, Talagala L. Safety considerations in MR imaging. Radiology 1990;176:593–606.[Abstract/Free Full Text]
4. Vanezis P. Non-compressive blunt impact injuries to the posterior compartment. In: Pathology of neck injury. London: Butterworth, 1989:92–112.
5. Hughes JT. Disorders of the spine and spinal cord. In: Adams JH, Duchen LW, editors. Greenfield's neuropathology (5th edn). London: Edward Arnold, 1992:1083–115.
6. Sances A, Myklebust J, Maiman D, Larson S, Cusick J, Jodat R. The biomechanics of spinal injuries. Crit Rev Bioeng 1984;11:1–76.
7. Fielding JW, Cochran GV, Lawsing JF III, Hohl M. Tears of the transverse ligaments of the atlas. A clinical and biomechanical study. J Bone Joint Surg Am 1974;56:1683–91.[Abstract/Free Full Text]
8. Sances A Jr, Weber RC, Larson SJ, Cusick JS, Myklebust JB, Walsh PR. Bioengineering analysis of head and spine injuries. CRC Crit Rev Bioeng 1981;5:79–122.[Medline]
9. Mertz H, Patrick L. Strength and response of the human neck. Proceedings of the 15th Stapp Car Crash Conference; 1971; San Diego, CA. New York: Society of Automotive Engineers, 1972:207.
10. Gadd C, Culver C. A study of the responses and tolerances of the neck. Proceedings of the 15th Stapp Car Crash Conference; 1971; San Diego, CA. New York: Society of Automotive Engineers, 1972:256–68.
11. Culver R, Bender M, Melvin J. Mechanisms, tolerances and responses obtained under dynamics of superior head impact, Final Report PB-299292. Ann Arbor, MI. University of Michigan Highway Safety Research Institute, 1978.
12. Patterson F. Instability of the upper cervical spine. J Bone Joint Surg Br 1973;55:456–7.
13. Davis F, Markley H. Rheumatoid arthritis with death from medullary compression. Ann Intern Med 1951;35:451.
14. Beighton P, Craig J. Atlanto-axial subluxation in the Morquio syndrome. Report of a case. J Bone Joint Surg Br 1973;55:478–81.
15. Martel W, Page J. Cervical vertebral erosions and subluxations in rheumatoid arthritis and ankylosing spondylitis. Arthritis Rheum 1960;3:546.
16. Storey G. Changes in the cervical spine in rheumatoid arthritis with compression of the cord. Ann Phys Med 1958;4:216.
17. Meijers KA, van Beusekom GT, Luyendijk W, Duijfjes F. Dislocation of the cervical spine with cord compression in rheumatoid arthritis. J Bone Joint Surg Br 1974;56:668–80.
18. Howcroft AJ, Jenkins DH. Potentially fatal asphyxia following a minor injury of the cervical spine. J Bone Joint Surg Br 1977;59:93–4.
19. Rana NA, Hancock DO, Taylor AR, Hill AG. Atlanto-axial subluxation in rheumatoid arthritis. J Bone Joint Surg Br 1973;55:458–70.

This Article Abstract Figures Only Full Text (PDF) Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Condon, B Articles by Hodgson, R Search for Related Content PubMed PubMed Citation Articles by Condon, B Articles by Hodgson, R