Impact of Imaging Landmark on the Risk of MRI-Related Heating Near Implanted Medical Devices Like Cardiac Pacemaker Leads

Transcription

1 BIOPHYSICS AND BASIC BIOMEDICAL RESEARCH - Note Magnetic Resonance in Medicine 65:44 50 (2011) Impact of Imaging Landmark on the Risk of MRI-Related Heating Near Implanted Medical Devices Like Cardiac Pacemaker Leads Peter Nordbeck, 1,2 * Oliver Ritter, 1 Ingo Weiss, 3 Marcus Warmuth, 2,4 Daniel Gensler, 2 Natalie Burkard, 1 Volker Herold, 2 Peter M. Jakob, 2,4 Georg Ertl, 1 Mark E. Ladd, 5 Harald H. Quick, 6 and Wolfgang R. Bauer 1 Implanted medical devices such as cardiac pacemakers pose a potential hazard in magnetic resonance imaging. Electromagnetic fields have been shown to cause severe radio frequency-induced tissue heating in some cases. Imaging exclusion zones have been proposed as an instrument to reduce patient risk. The purpose of this study was to further assess the impact of the imaging landmark on the risk for unintended implant heating by measuring the radio frequency-induced electric fields in a body phantom under several imaging conditions at 1.5T. The results show that global radio frequency-induced coupling is highest with the torso centered along the superior inferior direction of the transmit coil. The induced E-fields inside the body shift when changing body positioning, reducing both global and local radio frequency coupling if body and/or conductive implant are moved out from the transmit coil center along the z-direction. Adequate selection of magnetic resonance imaging landmark can significantly reduce potential hazards in patients with implanted medical devices. Magn Reson Med 65:44 50, VC 2010 Wiley-Liss, Inc. Key words: electromagnetic fields; cardiac pacemakers; equipment safety; medical implants; magnetic resonance imaging; MRI; RF heating INTRODUCTION Magnetic resonance imaging (MRI) in general poses relatively low patient risk compared to alternative imaging modalities, as it does not require the usage of ionizing radiation or iodated contrast agents. On the other hand, 1 Department of Internal Medicine I, University Hospital Würzburg, Julius- Maximilians-University, Würzburg, Germany. 2 Department of Experimental Physics V, Julius-Maximilians-University Würzburg, Würzburg, Germany. 3 Biotronik SE & Co. KG, Berlin, Germany. 4 Research Center Magnetic-Resonance-Bavaria, Würzburg, Germany. 5 Erwin L. Hahn Institute for Magnetic Resonance Imaging, University Duisburg-Essen, Essen, Germany. 6 Institute of Medical Physics, Friedrich-Alexander-University Erlangen- Nürnberg, Erlangen, Germany. Grant sponsors: Bayerische Forschungsstiftung (Bavarian Research Foundation), Deutsche Gesellschaft für Kardiologie (German Cardiac Society). *Correspondence to: Peter Nordbeck, M.D., Medizinische Klinik und Poliklinik I, Kardiologie/Elektrophysiologie, Universitätsklinikum Würzburg, Oberdürrbacher Str. 6, Würzburg 97080, Germany. nordbeck_p@ medizin.uni-wuerzburg.de Received 6 January 2010; revised 13 July 2010; accepted 13 July DOI /mrm Published online 30 August 2010 in Wiley Online Library (wileyonlinelibrary.com). VC 2010 Wiley-Liss, Inc. 44 increasing numbers of patients with implanted passive and/or active electrically conducting medical devices raise safety problems in daily routine, as additional procedural risks emerge when performing MRI in these patients. Increasing clinical relevance of both MRI in general and implanted medical devices make MR safety in these patients an increasingly relevant issue. Recently, the first cardiac pacemaker officially labeled MR conditional has been introduced (1). When considering MRI in patients with cardiac pacemakers, particularly safety concerns regarding pronounced radio frequency (RF)-induced tissue heating at the tip of the pacemaker lead are the focus of the current discussion. Recent investigations have shown the general potential for severe heating to take place in a clinical setup, e.g., reporting significant pacing threshold increases or elevated serum troponin after MRI in pacemaker patients (2,3), and even permanent negative events have been reported (4,5). Therefore, device manufacturer guidelines for MR-labeled products include certain restrictions to ensure patient safety. Such guidelines might include restrictions regarding imaging landmark e.g., excluding cardiac or spinal MRI in the newly introduced MR-conditional pacemaker device mentioned above (1,6). Preliminary in vivo investigations indeed suggest the risk for unintended implant heating to be diminished if the implant is moved out of the imaging region (7). Our group recently showed that unintended implant heating is strongly dependent on the MRI-induced E-fields inside the body, and, moreover, the implant geometry and positioning inside the body, making unintended implant heating particularly likely if the implant follows the course of the E-fields induced inside the body (8,9). The aim of this study was to investigate systematically the impact of body positioning/imaging landmark inside the scanner bore (or, more specifically, inside the RF transmit coil) on the spatial distribution of RF energy deposition as a mandatory prerequisite for implant heating, and to clarify whether patient safety can be significantly increased by specification of imaging exclusion zones in patients with implanted conductive devices like cardiac pacemakers. Multiple MR scanners were assessed to allow for an estimation of how different hardware (B1- field, RF transmit coil, etc.) and software (pulse sequence, SAR calculation, etc.) properties influence RFinduced heating dependent on imaging landmark.

2 MR-Induced Implant Heating 45 METHODS All measurements were performed at 1.5 T in an acrylic plastic head/torso phantom according to the ASTM standard F a [ cm 3, (10)] filled with 45 kg of a saline fluid providing conductivity similar to that of body tissue as previously proposed (10) (0.47 S/m at 64 MHz, roughly matching brain tissue, and at the same time yielding a compromise for different body tissues and fluids). The phantom was registered supine, head first, with a patient weight of 45 kg. Three different 1.5 Tesla MR scanners were used: a Siemens Magnetom Vision (Scanner 1), a Siemens Magnetom Avanto (Scanner 2, both Siemens Healthcare, Erlangen, Germany), and a Philips Gyroscan Intera (Scanner 3, Philips Healthcare, Best, The Netherlands). The integrated RF body coil was used for pulse excitation and signal reception in all scanners. For the initial measurements, the geometric center of the phantom torso was positioned in the isocenter of the scanner bore (x, y, and z-direction), later being moved to different imaging landmarks along the bore axis by MR table offset. A three-dimensional map of MRI-related E-field distribution inside the phantom fluid was then sampled during MR imaging in several setups using a custom E-field measurement system. The general measurement setup has been described in detail in an earlier publication (8). This measurement system is composed of a short dipole sensor driving a light emitting diode (LED), which converts the RF-induced electric signal into light intensity. The light signal from the LED is fiber-optically transmitted outside the scanner room, reconverted, amplified, and finally measured with a digital multimeter. The LED does not flicker with the lamor frequency of 64 MHz, but follows the envelope of the RF signal (which is usually a sinc pulse). The root mean square value of this RF pulse envelope is then assessed using a true RMS measurement instrument. This measured value is proportional to the local magnitude of the RF-induced E-fields; however, the proportionality factor is sequence dependent, as it depends on the shape and time schedule of the RF pulses. Therefore, the same pulse sequence should be used for direct comparability of the measurements results and the results have to be normalized to a reference (maximum value) to ensure that this proportionality factor does not influence the measurement results aimed at, which were the relative change of energy deposition as a function of imaging landmark in the current study. This approach allows for precise, systematic, and reasonably fast determination of local E-field magnitudes induced by MRI inside the body, which has been proven to closely correlate with resultant implant and, therefore, tissue heating (9). Our sensor proved not to respond to E-field magnitudes induced by gradient fields. Therefore, the E-field measurement results relate to RF only. The E-field distribution inside the phantom was determined in each scanner by shifting the E-field probe on a measurement grid inside the phantom fluid while running a turbo spin echo pulse sequence. This measurement grid consisted of 168 measurement points equally distributed over the fluid middle layer of head and torso. Both z-directional and x-directional E-field components were acquired separately for each measurement point, using a short straight E-field probe. Y-directional E-field components have been shown to be negligible in this setup (8), and, therefore, were not acquired. RF-transmit power during the measurements was set to a scannerindicated time averaged power of 100 W by adjustment of pulse sequence flip angle and repetition time. Local E-field strength was later scaled to normalized units, dependent on the scanner-specific E-field maximum, to exclude interscanner variations caused by differences in pulse sequence or scanner SAR calculation algorithms. In each scanner, the E-field distribution inside the phantom medium was determined for several imaging landmarks, starting with the phantom torso centered at the isocenter (z-axis: 0 cm) and later moved along the z-axis in 10 cm increments (Fig. 1). For this purpose, the phantom was moved by table offset, and an E-field map acquired for each respective landmark while running the described pulse sequence and moving the E-field sensor oriented in x- and z-direction over the measurement grid. All E-field measurement values are reported in percentage normalized to the scanner specific RF-induced E-field maximum, as the focus of the study was on relative changes under varying conditions rather than absolute measures, which are known to be very difficult to compare when using different pulse sequences in multiple scanners. The normalized (arbitrary) values allow for a direct comparison of relative E-field amplitude changes under various conditions, independent on pulse sequence or scanner in use. RESULTS With the phantom centered in the head-foot direction within the scanner bore and, therefore, transmit coil, the general E-field distribution was found to show similar patterns in all three scanners: E-field lines and maximum E-field strength were generally found to follow the contours of the phantom, with particularly high field strength in z-direction near the lateral phantom walls at the center of the scanner bore. Moving the imaging landmark and phantom along the z-axis in the head or hip direction resulted in an E-field distribution inside the phantom shifted in the same direction as the imaging landmark. A representative overall E-field distribution for different imaging landmarks at the medium liquid layer (y ¼ 0 cm) is shown for Scanner 2 in Fig. 2, with blue color representing areas with lowest and red color representing areas with highest local E-field strength. As apparent in the images, the E-field magnitude maximum roughly follows the imaging landmark, e.g., towards the head if the head is imaged. Although these shifted E-field patterns in general were found in all three scanners, remarkable differences were found between the scanners regarding the extent of E-field shift and E-field strength reduction under changing imaging conditions. Differences become particularly apparent comparing the spatial z-directional E-field components along the lateral phantom walls under several imaging conditions (imaging landmarks). The results of the measurements for all three scanners are condensed

3 46 Nordbeck et al. FIG. 1. Experimental setup. A: Measurement grid along the phantom filling. Each square represents one measurement point. Dotted line: measurement line along left phantom wall (-> see Figure 4). B: The geometric center of the phantom torso liquid was centered inside the scanner bore and E-fields were measured using the custom E-field probe. E-field maps were then acquired for various imaging landmarks adjusting the MR table offset, and therefore phantom position, along the z-direction. Accordingly, the imaging landmarks reflect the difference between scanner isocenter and phantom torso isocenter. Partial E-field distributions in different scanners are compared in Figs. 3 and 5 for the results of the measurement grid along the lateral phantom wall. in Fig. 3. Respective E-field strength is given in normalized units (percent), referring to the maximum measured value. As apparent in the graphs, in all three scanners the highest overall E-field strength was found with both the phantom centered inside the transmit coil and the E- field sensor placed near the center of the transmit coil in z-direction along the walls inside the phantom. Likewise, all three scanners showed a similar shift of E-field distribution in the respective direction if the phantom was moved either in the head ( ) or feet (þ) direction, with feet-head-specific differences mainly concentrated in the head section and at the wall of the phantom toward the feet. In Scanner 1, shifting the imaging landmark to þ10 cm reduced peak E-field strength in the phantom torso by 23%, at þ20 cm by 37%, at þ30 cm by 45%, at þ40 cm by 69%, and at þ50 cm by 84% of the maximum. In Scanner 2, imaging landmark dependent reduction of the E-field maximum in the phantom torso was only 1.1% at þ10 cm, 11% at þ20 cm, 48% at þ30 cm, 76% at þ40 cm, >99% at þ50 cm, and in Scanner 3 1.0% at þ10 cm, 8% at þ20 cm, 29% at þ30 cm, 67% at þ40 cm, and >99% at þ50 cm (Fig. 4). Comparing the E-field in the three different scanners at imaging landmark z ¼ 0 cm, Scanner 1 and Scanner 2 showed a very similar E-field distribution, while Scanner 3 revealed a broader distribution curve with steeper decrease to the phantom ends (Fig. 5). Shifting the imaging landmark toward either end of the phantom decreased peak heating in all scanners, but to a different extent: at the imaging landmark of þ40 cm, the position of peak heating was shifted by only 14 cm in the direction of the imaging landmark in Scanner 1, but by 34 cm in Scanner 2, and 33 cm in Scanner 3. Overall peak heating value in the phantom torso decreased quickly over the first centimeters of imaging landmark shift away from the phantom center in Scanner 1, and only slowly over the first centimeters of imaging landmark shift in Scanner 2 and 3 (Fig. 5). DISCUSSION We investigated energy deposition in MRI under changing imaging conditions as a correlate for the relative risk of RF-induced implant heating, particularly focusing on the impact of varying imaging landmarks. Amongst other measures, moving the imaging landmark away from a conductive implant in the body has recently been proposed to reduce the risk for heating (7), but these investigations were performed in vivo, where many unpredictable factors make generalizations on this aspect extremely difficult. Therefore, recommendations on specific imaging exclusion zones, as for example given for a recently introduced cardiac pacemaker system labeled MR conditional (1), are currently difficult to generalize. The problem of MRI-related heating near conductive active or passive devices is a very complicated issue, with a very large number of contributing factors determining the ultimate risk for heating in a given situation. As an example, this has become apparent by publication of a case report describing permanent neurological deficit due to severe MRI-related electrode heating in a patient with a neurostimulator (5). This case is particularly interesting, as an MR conditional device was used, which has probably been used in many MRI investigations of neurostimulator patients without complications when following the device specific guidelines. Cranial MRI is assumed to be relatively safe in these patients compared to other MR investigations if specific precautions like restrictions in SAR and use of a transmit-andreceive head coil are followed (5) even though there are

4 MR-Induced Implant Heating 47 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi FIG. 2. E-field distribution Ex 2 þ Ez 2 in Scanner 2 (Siemens Avanto) dependent on imaging landmark. Imaging landmark is varied from (A) 30 cm (hip) to I: þ50 cm (brain). Relative local E-field strength (in percent, normalized to the maximum value) is color encoded. While overall E-field strength decreases if the phantom is moved away from the center position, local E-field strength especially near the imaging landmark might increase, e.g., in the head section with a table offset of þ40 cm compared to an imaging landmark in the torso center. still few data published providing extensive insights in these particular aspects regarding RF-related heating of medical implants. Likewise, many investigators have suggested the risk for unintended implant heating to be increased with increasing MR field strength. The aforementioned case report demonstrates profoundly that given the complexity of the subject even though such simplifications may not be inaccurate, they should not be uncritically generalized neglecting additional contributing factors. We, therefore, now provide a systematic investigation of the dependency of the E-field distribution on imaging

5 48 Nordbeck et al. FIG. 3. Relative z-directional E-field strength and E-field distribution along the left lateral phantom wall (dashed arrow Fig. 1A) dependent on imaging landmark for three different scanners. X-axis: point of measurement inside the phantom along the phantom wall. Y-axis: relative E-field strength, normalized to the scanner dependent E-field maximum. Different colors represent different imaging landmarks ( 30 to þ50 cm relative to the imaging landmark at torso center). has the benefit that disturbances in signal transmission can be avoided. On the other hand, LEDs show a nonlinear activation behavior, meaning that the sensor components have to be adjusted to the electric quantity to be measured (or vice versa) to provide a nearly linear characteristic within the intended working range (8,9). In this case, the output of the measurement amplifier is expected to be proportional to the electric current intensity through the LED (with the square of the electric current intensity, as well as the square of the induced voltage, being proportional to the electric power converted into heat, and, therefore, being proportional to resultant temperature changes at device-to-tissue interfaces). Once the sensor characteristics have been adjusted and investigated, the appropriate working range for a given sensor should be maintained by not exceeding the predefined saturation cutoff values. The working range for a given sensor can be varied further through variations in scanner RF output power, which has been shown to be directly proportional to the RF-induced heating (13), and, therefore, can be used to adapt the sensor working range to various conditions if needed. There are two main effects that the data show: (1) if the implant is far away from the imaging landmark, then there is little potential for heating, and (2) if the landmark is moved away from the phantom middle (i.e., away from the middle of the patient, where there is the most cross-section to induce eddy currents in the tissue), then the global heating risk is also lower, due to a highly significant general reduction in the overall E-field strength. As shown in the current study, the specific risk reduction is to some extent scanner-dependent, which is in line with earlier findings of other groups (14). We found a mean reduction of the peak E-field strength by 40% in the phantom torso, roughly corresponding to a reduction in heating risk to one-third if other factors are excluded from this estimation. On the other hand, the current results show that the E-field peak in the head section of the phantom can be increased if the imaging landmark is moved from the torso toward the head (Fig. 2). A general tendency of landmark in a standardized phantom. These RF-induced E-fields have already been shown to be the direct cause of MRI-related implant heating (8). Translating the results to the problem of implant heating, it has to be kept in mind that there is a quadratic relation between the relative E-field strength and resultant implant heating (9), implying a fourfold increase in implant heating if E-field strength is doubled [provided an otherwise identical setup (9,11,12)]. Likewise, the absolute risk for heating is reduced to a quarter if the results of E-field measurements show a reduction of 50%. The E-field measurement procedure used in the current study is based on converting local electric signal into light signal using a light emitting diode (LED). This FIG. 4. Reduction of E-field strength maximum in the phantom torso dependent on the imaging landmark. Maximum overall E- field values were found at imaging landmark 0 cm in all three scanners. Landmark dependent reduction in phantom torso peak E-field value is shown for each scanner.

6 MR-Induced Implant Heating 49 FIG. 5. Relative E-field strength and E-field distribution in three different MR scanners along the left lateral phantom wall (dashed arrow Fig. 1A) for various imaging landmarks (same data as Fig. 3, displayed for comparison between scanners; A: 0 cm to F: þ50 cm). Light gray: Scanner 1 (Siemens Vision), dark gray: Scanner 2 (Siemens Avanto), black: Scanner 3 (Philips Intera). X-axis: point of measurement inside the phantom along the phantom wall. Y-axis: relative E-field strength, normalized to the scanner specific maximum value. With the phantom centered in the scanner bore (isocenter imaging landmark: liver), E-field distribution is very similar in all three scanners, varying incrementally with increasing table offset (z þ30 cm imaging landmark: neck). maximum power deposition to remain in the region close to the imaging landmark was found in all three scanners, although to a different extent. In our opinion, the general findings of other groups in the past suggesting cranial MRI to be relatively unproblematic with regard to the risk for implant heating (15,16) results in large part from the generally lower overall scanner transmit power used in cranial MRI, which in turn generates lower overall E-fields. Direct comparison of the three different scanners investigated in the current study revealed that Scanner 1 showed relatively little decrease in E-field strength if the imaging landmark was placed on peripheral regions of the phantom. Scanner 2 and 3, on the other hand, exhibited E-field characteristics similar to one another under varying imaging landmarks, including a falloff in RF coupling for peripheral landmark positions. These measurement results most likely reflect the differently sized RF body coils, with Scanner 1 representing a long-bore system equipped with a 1000 mm long RF body coil, while Scanner 2 and 3 were short-bore systems with an RF coil length of only 600 mm, as approximated from loop

7 50 Nordbeck et al. receiver RF transmission measurements inside the different scanners. This has important implications on peripheral RF coupling. With the phantom largely moved outside the RF coil, RF coupling strongly decreased for peripheral imaging landmarks in Scanner 2 and 3, while Scanner 1 provided much broader energy transmission characteristics for peripheral imaging landmarks, presumably since the phantom was still partially inside the RF coil, as visible in the E-field magnitudes shown in Figs. 3 and 5. All measurements in the present study were performed in a phantom conforming to ASTM standard F a (10). The general findings can, therefore, be translated to other conditions relatively easily, as unpredictable disturbing factors are excluded in this setup. However, a direct extension of the results of the current study to the situation in the human body without taking into account the other contributing factors (17) is invalid. That is, a shift of imaging landmark of þ30 cm will result in a different E-field shift in the human body, amongst other factors caused by body shape and tissue inhomogeneities, even though the general RF power deposition principles remain the same as shown in the current study. CONCLUSION We investigated RF-related E-field distribution as a correlate for potential unintended heating of electrically conductive implants in the body using various imaging landmarks in three different MR scanners. The results show that overall E-field magnitudes strongly decrease if the RF coupling is reduced, for example by moving the landmark toward the ends of the object being imaged. Due to this fact and a highly significant decrease in local E-field strength with increasing distance from the imaging landmark with the exact E-field distribution being scanner dependent the overall potential of implanted devices to develop severe RF-related heating is also reduced with increasing distance from the imaging landmark. If MRI of patients with implanted electrically conductive medical devices is considered, adequate selection of imaging landmark is, therefore, an important component to be taken into account to ensure patient safety. REFERENCES 1. Sutton R, Kanal E, Wilkoff BL, Bello D, Luechinger R, Jenniskens I, Hull M, Sommer T. Safety of magnetic resonance imaging of patients with a new Medtronic EnRhythm MRI SureScan pacing system: clinical study design. Trials 2008;9: Roguin A, Zviman MM, Meininger GR, Rodrigues ER, Dickfeld TM, Bluemke DA, Lardo A, Berger RD, Calkins H, Halperin HR. Modern pacemaker and implantable cardioverter/defibrillator systems can be magnetic resonance imaging safe: in vitro and in vivo assessment of safety and function at 1.5 T. Circulation 2004;110: Sommer T, Naehle CP, Yang A, Zeijlemaker V, Hackenbroch M, Schmiedel A, Meyer C, Strach K, Skowasch D, Vahlhaus C, Litt H, Schild H. Strategy for safe performance of extrathoracic magnetic resonance imaging at 1.5 tesla in the presence of cardiac pacemakers in non-pacemaker-dependent patients: a prospective study with 115 examinations. Circulation 2006;114: Wright O. Patient killed by heart blunder, The Times, April 3, Henderson JM, Tkach J, Phillips M, Baker K, Shellock M, Rezai AR. Permanent neurological deficit related to magnetic resonance imaging in a patient with implanted deep brain stimulation electrodes for Parkinson s disease: case report. Neurosurgery 2005;57: EnRhythm MRI SureScan User Manual. Minneapolis: Medtronic; Luechinger R, Zeijlemaker VA, Pedersen EM, Mortensen P, Falk E, Duru F, Candinas R, Boesiger P. In vivo heating of pacemaker leads during magnetic resonance imaging. Eur Heart J 2005;26: Nordbeck P, Fidler F, Weiss I, Warmuth M, Friedrich MT, Ehses P, Geistert W, Jakob PM, Ladd ME, Quick HH, Bauer WR. Spatial distribution of RF-induced E-fields and implant heating in MRI. Magn Reson Med 2008;60: Nordbeck P, Weiss I, Ehses P, Ritter O, Warmuth M, Fidler F, Herold V, Jakob PM, Ladd ME, Quick HH, Bauer WR. Measuring RF-induced currents inside implants: impact of device configuration on MRI safety of cardiac pacemaker leads. Magn Reson Med 2009;61: American Society for Testing and Materials. Standard test method for measurement of radio frequency induced heating near passive implants during magnetic resonance imaging (F a). West Conshohocken, PA: ASTM International; p. 11. Armenean C, Perrin E, Armenean M, Beuf O, Pilleul F, Saint-Jalmes H. RF-induced temperature elevation along metallic wires in clinical magnetic resonance imaging: influence of diameter and length. Magn Reson Med 2004;52: Mattei E, Triventi M, Calcagnini G, Censi F, Kainz W, Mendoza G, Bassen HI, Bartolini P. Complexity of MRI induced heating on metallic leads: experimental measurements of 374 configurations. Biomed Eng Online 2008;7: Nordbeck P, Bauer WR, Fidler F, Warmuth M, Hiller KH, Nahrendorf M, Maxfield M, Wurtz S, Geistert W, Broscheit J, Jakob PM, Ritter O. Feasibility of real-time MRI with a novel carbon catheter for interventional electrophysiology. Circ Arrhythm Electrophysiol 2009;2: Baker KB, Tkach JA, Phillips MD, Rezai AR. Variability in RFinduced heating of a deep brain stimulation implant across MR systems. J Magn Reson Imaging 2006;24: Soletra Neurostimulator for Deep Brain Stimulation, Model 7246, Physician and Hospital Staff Manual. Minneapolis: Medtronic, Rezai AR, Phillips M, Baker KB, Sharan AD, Nyenhuis J, Tkach J, Henderson J, Shellock FG. Neurostimulation system used for deep brain stimulation (DBS): MR safety issues and implications of failing to follow safety recommendations. Invest Radiol 2004; 39: Levine GN, Gomes AS, Arai AE, Bluemke DA, Flamm SD, Kanal E, Manning WJ, Martin ET, Smith JM, Wilke N, Shellock FS; American Heart Association Committee on Diagnostic and Interventional Cardiac Catheterization; American Heart Association Council on Clinical Cardiology; American Heart Association Council on Cardiovascular Radiology and Intervention. Safety of magnetic resonance imaging in patients with cardiovascular devices: an American Heart Association scientific statement from the Committee on Diagnostic and Interventional Cardiac Catheterization, Council on Clinical Cardiology, and the Council on Cardiovascular Radiology and Intervention: endorsed by the American College of Cardiology Foundation, the North American Society for Cardiac Imaging, and the Society for Cardiovascular Magnetic Resonance. Circulation 2007;116: