MR Imaging-Related Heating of Deep Brain Stimulation Electrodes: In Vitro Study

Methods

In Vitro Model

In vitro testing was performed on a 1.5-T MR imaging system (Vision software, version Numaris 3.0; Siemens Medical Systems, Iselin, NJ) with a head transmit-receive coil. A special Plexiglas phantom designed to approximate the size and shape of the human head and torso was developed and filled with a semi-solid aqueous gel prepared to simulate the electrical conductivity and thermal convection properties of tissue (26). The weight of the filled phantom is 70 lb and is the weight entered at registration for purposes of SAR calculation. A plastic grid frame with adjustable posts was placed at the bottom of the phantom to allow consistent positioning and support of the neurostimulators, extensions, and leads within the phantom (Fig 1). Currently, the only Food and Drug Administration-approved device for chronic DBS is the Activa Tremor Control System (Medtronic, Inc., Minneapolis MN). This neurostimulation system includes the model 7426 Soletra implantable pulse generator, the model 7495 quadripolar extensions, and models 3387 and 3389 DBS leads. In this study, we positioned bilateral DBS electrodes, extensions, and implantable pulse generators in a manner to emulate the clinical method of subcutaneous pectoral placement of the implantable pulse generators and subcutaneous extension from the implantable pulse generator up the chest, along the neck, and over the cranial convexity to connect with the electrode lead, which passes through the cerebral cortex, with electrode contact plates in the subthalamic region (Fig 2). The length of the DBS lead is 40 cm, and it is insulated along its entire length, except for the connector and the four contact plates in the distal 10 mm of the lead. It is desirable to keep the extension connector in a stable, nonmoving position at the level of the cranium and not in the neck. Therefore, it is the most common current clinical practice to loop the excess electrode lead around the burr hole and then connect to the extension. Similarly, the extensions are 51 cm. Frequently, excess length is looped around the perimeter of the implantable pulse generator. A previous study (26) showed that this was the configuration that produced the greatest temperature elevations when using the head transmit-receive coil. In that same study, the worst case scenario was to use the body transmit coil and take the entire excess length of electrode and extension and coil it in the pocket of the implantable pulse generator (26), a configuration that was associated with ≤25.3°C elevation in temperature (at 3.9 W/kg whole body averaged SAR). Thus, for these in vitro experiments, one lead (right) was configured in the typical, worst case configuration for head transmit-receive coils and one lead (left) was configured in the worst case configuration based on body transmit coils, for simultaneous heating tests. The electrodes are positioned parallel and coronally coplanar, spaced 3 cm apart. Each electrode system is electronically isolated, and pilot studies of unilateral versus bilateral placements have shown no interdependence for bilateral electrodes in these and several other electrode configurations.

• Download figure
• Open in new tab
• Download powerpoint

Fig 1.

Phantom and experimental setup. Photograph shows the plexiglass phantom before filling with the gel saline solution shows the plastic grid frame and adjustable posts, which allow precise positioning and support of the dual implantable pulse generators, extensions, and DBS leads. The cables for the four fluoroptic thermometry probes are seen coursing off the bottom of the photograph.

• Download figure
• Open in new tab
• Download powerpoint

Fig 2.

DBS lead mean temperature elevation as a function of local SAR. Plot shows greatest observed mean temperature elevation for 1 min of data collection, obtained at the time of peak heating during the 10- to 15-minute imaging time interval, versus local SAR value for single and multisection fast spin-echo images. Linear regression (all data points): r = 0.994, P < .001, x variable = 0.88, and y intercept = −0.13.

Results

The temperature elevations observed with the standardized fast spin-echo measurements over the range of SAR values are listed in the Table. We found virtually identical, highly linear, statistically significant relationships between the local averaged and whole body averaged SAR values and the observed temperature increase measured at the DBS electrode tips for the single section and multisection fast spin-echo sequences (Fig 2; linear regression, r = 0.994; P < .001; x variable = 0.88; y intercept = −0.13). A maximum temperature elevation of 6.7°C was observed with the sequence having a local SAR of 7 .3 W/kg (0.24 W/kg whole body averaged). The electrode configuration that was associated with the greatest MR imaging-related heating in the head transmit-receive coil was the right lead (excess lead coiled at the burr hole, excess extension coiled at the implantable pulse generator), confirming the findings of a previous study (26). By interpolation, the SAR values at which a 1°C temperature rise at the electrode tips is expected is approximately 1.2 W/kg local or approximately 0.045 W/kg whole body averaged and the SAR values at which a 2°C temperature rise at the electrode tips is expected is approximately 2.4 W/kg local or approximately 0.09 W/kg whole body averaged. The background temperature increase was <0.5°C for all measurements, even at high SAR values.

The temperature elevations observed with the clinical imaging sequences are shown in the Table. The fast spin-echo, echo-planar, and magnetization prepared rapid acquisition gradient-echo imaging sequences had local SAR values <0.5 W/kg. None of these clinical imaging sequences was associated with DBS electrode temperature elevations >0.5°C. The experimental temperature elevations observed with fast spin-echo, echo-planar, and magnetization prepared rapid acquisition gradient-echo imaging sequences were all within 0.1°C of that predicted by the regression analysis above. Therefore, this in vitro model indicates that fast spin-echo, echo-planar, and magnetization prepared rapid acquisition gradient-echo imaging sequences performed with typical parameters and within these SAR restrictions are associated with no significant MR imaging-related heating. The echo-planar imaging sequences, which have very low SAR values, showed no significant heating of the DBS electrodes above the reference probe, indicating that the high dB/dt associated with the echo-planar imaging sequence has no apparent heating effects independent of the SAR considerations.

Quantitative magnetization transfer imaging and magnetization transfer-suppressed MR angiography, with local SAR values of 3.4 and 4.3 W/kg, respectively, had temperature elevations of <2.0°C. Interestingly, the temperature elevations observed with both of these sequences, with which most the RF deposited in a long duration, off resonance magnetization transfer prepulse, was less than expected based on our previous results by using the standardized fast spin-echo sequences.

Discussion

MR imaging-related heating at the DBS electrode tips occurs when the RF field induces a current in the lead and extension wires and some of this induced current passes through the electrode contact plates and into the surrounding tissues. The temperature increase at the electrode contacts is proportional to the square of the emanating current. Because the SAR is proportional to the square of the induced electric field, it is expected, and this study confirmed, that a linear relationship exists between the SAR and the temperature elevation measured at the contact plates (7, 19). Regression analysis of the experimental data also confirmed the expectation that the y intercept should be essentially zero (ie, with no applied RF, there is no temperature elevation). It is important to recognize that the RF induction of current in the electrode lead is a very different phenomenon from what might be understood about RF excitation signal intensity generation in MR imaging. The current induced in the conductor depends mainly on its electrical conductivity properties, the length and configuration (loops) of the conductor in the primary RF field, its position in the primary RF field (there are inhomogeneities in the RF profile, especially adjacent to coil elements and in the periphery of the RF field), and the effective RF power. The application of a section-select gradient produces only a small change (a few hundred hertz) in the local carrier frequency of the RF (ie, the Larmor frequency, or 63.5 MHz for this study). This has no significant impact on the induction of current in the lead. When RF is applied, it is always through the entire sensitive volume of the transmit coil. Theoretically, there should be no significant dependence of the induced current in the lead (or resultant heating) on section-select events (ie, single section, multisection, nonselective, imaging orientation, positioning an imaging section on the lead or through the lead, etc.). The concordance of the single section-multisection fast spin-echo data reported herein supports these concepts about MR imaging-related heating of conductors.

The phantom design, sequence design, electrode lead positioning, and use of a physiological gel saline solution were chosen to emulate a worst case MR imaging-related heating scenario, in which MR imaging-related heating is maximized, thermal convection is constrained (compared with pure aqueous solutions), and no tissue perfusion occurs to carry away heat. Using this worst case scenario in our experimental design maintains a conservative approach to the inference of “safety” when performing certain MR imaging of patients with implanted DBS electrodes. Previous studies of RF and other thermal ablation techniques have shown that reversible thermal lesions occur when the local temperature is elevated to the 42 to 44°C range (a 5–7°C elevation over the normal body temperature of 37°C) and that irreversible thermal lesions occur when the local temperature is elevated above 45°C (>8°C elevation over normal body temperature) (27–30). Febrile illnesses associated with fever to the 40°C range are not associated with any evidence of neural tissue damage (31). Therefore, transient temperature elevations of 1 to 2°C for several minutes while undergoing MR imaging examinations are expected to be easily tolerated with no adverse physiological effects. As is evidenced in several recent articles about RF thermal ablation of liver lesions, local tissue perfusion has a dramatic effect on heat deposition (30, 32–36). Normal brain perfusion would diminish the temperature elevations associated with MR imaging compared with the results of this in vitro work; however, participant safety, experimental design issues with electrode and fluoroptic probe positioning, and research ethics preclude trying to perform in vivo testing of MR imaging-related heating associated with DBS electrodes in human study participants. An unpublished survey of functional neurosurgical investigators at a recent meeting indicates that, to date, several hundred MR imaging studies of patients with this particular neurostimulation system have been performed with no adverse events (A.R. Rezai, unpublished results). Thus, this in vitro work provides confirmation that conventional clinical MR imaging sequences of the brain, performed with a head transmit-receive coil, are associated with minimal, physiologically tolerable, safe temperature elevations at the DBS electrode tips.

A second mechanism by which an electrical current can be induced in a conductor is by fluctuating magnetic field gradients; therefore, there was some concern that the high dB/dt (approximately 70 T/s) echo-planar imaging sequences may be associated with some additional heating beyond that associated with RF transmission. We found this not to be the case; based on the local SAR for the echo-planar imaging sequence, we would have predicted a temperature elevation of 0.3°C, which was exactly our measured result. Therefore, one can assume that for clinical MR imaging sequences, MR imaging-related heating is dependent only on RF electrical field induction of current in the leads.

A somewhat unanticipated result was that the temperature elevations associated with two MR imaging sequences using magnetization transfer were less than expected for their relatively high SAR values. Both of these magnetization transfer imaging methods use a long duration, high amplitude, off resonance magnetization transfer pulse. These results should be interpreted with extreme caution, because we expected to observe temperature elevations in the 4°C range based on the SAR values of the magnetization transfer imaging sequences, which would be unacceptable from an MR imaging safety standpoint. Although there are numerous magnetization transfer pulse parameters and tissue-specific issues that influence magnetization transfer imaging techniques (37–39), theoretically, there should be no significant dependence of the amount of induced current in the electrode lead related to the magnetization transfer pulses being a few kilohertz off resonance. Likewise, the RF profile of the head coil is not going to change because the transmitted RF is off resonance. Calculations of SAR values may be different for various manufacturers and models and may even be dependent on the specific RF pulses used in a particular sequence. The MR imaging system used in this trial does not make any modifications of SAR calculations based on tissue distributions or based on the magnetization transfer pulses being off resonance or having slightly different tissue RF absorption. Its SAR calculations are based on the summation of the time integrals of the RF pulses, the sequence TR, and the entered patient weight. The most likely explanation for the less-than-expected observed heating with the magnetization transfer imaging sequences relates to the MR system being an older clinical imaging system that uses an RF amplifier with tubes, not solid state circuitry. For the long duration, high amplitude magnetization transfer pulses, which have flip angle equivalents of 700 to 1200 degrees, the RF amplifier cannot maintain the voltage necessary for the high RF output and powers down to prevent tube overload. Thus, with these magnetization transfer imaging sequences, the machine is not actually transmitting the full RF power that the SAR calculation would represent. This hardware-related phenomenon is likely to be machine, model, and even individual imager specific and will likely vary over the life of the RF amplifier tubes. This effect will be magnetization transfer imaging sequence dependent and is also likely to be patient dependent, varying with the loading of the coil, because transmitter adjustments will determine how much voltage is needed to produce 90, 180, and all other degree RF pulses. Therefore, a great deal of uncertainty exists with respect to accepting as representative the apparently minor temperature elevations (1.5–1.8°C) seen with the quantitative magnetization transfer and magnetization transfer-suppressed MR angiography sequences. Most clinical quantitative magnetization transfer imaging sequence techniques are designed to have moderately high SAR values, in the 3.0 to 4.0 W/kg local SAR range or higher, because increasing the effective magnetization transfer power increases the dynamic range of the magnetization transfer ratio values (37). It is expected that maintaining SAR values of approximately 2.0 W/kg local with the magnetization transfer imaging sequences would be associated with minor, tolerable temperature elevations, but we have not performed these confirmatory experiments. Therefore, in keeping with a conservative approach to MR imaging safety, we recommend avoiding sequences using magnetization transfer imaging for patients with neurostimulation systems until more detailed testing resolves these uncertainties.

We have shown that MR imaging-related heating for fast spin-echo, gradient-echo, and echo-planar imaging sequences (and, by implication, for conventional spin-echo sequences as well) in the head transmit-receive coil is linearly correlated with SAR and that sequences performed within the prescribed SAR range of ≤2.4 W/kg local (0.09 W/kg whole body averaged) should be safe from a standpoint of MR imaging-related heating. This is a very important result from the standpoint of potential clinical imaging scenarios for patients with DBS electrodes. It is possible to safely obtain high resolution stereotactic 3D gradient-echo images and T2-weighted fast spin-echo images for documenting electrode contact plate positioning in patients with poor or worsening response to DBS and for placement of additional DBS electrodes. MR images to evaluate tumors and strokes, contrast-enhanced images, diffusion- and perfusion-weighted images, and non-magnetization transfer MR angiograms can all be safely obtained. Perhaps most importantly for future neuroimaging research in the functional and restorative neuroscience field, functional MR imaging studies can be performed safely from an MR imaging-related heating standpoint. It is acknowledged that other specific clinical sequences may ordinarily be performed with SAR values above the suggested limits, but simply prolonging the TR appropriately should allow these to be performed safely for patients with DBS electrodes.

There are several other MR imaging safety issues that were not addressed in this investigation, such as inappropriate neural stimulation, dislodgement-translation of the device by the static magnetic field, and disruption of the functional integrity of components. In general, previous studies have shown that if the implantable pulse generator is programmed to “off” and the voltage is set to 0 volts, no electrical stimulation occurs and the device is not damaged (1). Static magnetic field interactions, such as displacement or torque of the leads, are a subject of current investigation.

Food and Drug Administration guidelines (1998) define MR imaging operating conditions that may pose significant risk to patients. For RF energy, significant risk exposures are associated with an SAR >4.0 W/kg whole body for 15 min, 3.0 W/kg averaged over the head for 10 minutes, 8.0 W/kg in any gram of tissue in the head or torso for 15 minutes, or 12 W/kg in any gram of tissue in the extremities for 15 minutes (7, 11, 40). It is important to recognize that our recommendations for safe RF exposures in patients with DBS electrodes (<2.4 W/kg local SAR over the head, 0.09 W/kg whole body averaged) are stricter than the current Food and Drug Administration guidelines and apply only to imaging of the brain with a head transmit-receive coil. Previous Food and Drug Administration guidelines (12) had considered there to be no significant risk associated with RF exposure when MR imaging produced <1°C measured temperature elevation in various body parts or skin surfaces. In this manner, brain MR imaging with head transmit-receive coils in patients with implanted DBS electrodes should pose no significant risk with local SAR values <1.2 W/kg (0.045 W/kg whole body averaged). Minor, physiologically tolerable, safe temperature elevations in the 1 to 2°C range are expected with local SAR values ≤2.4 W/kg (0.09 W/kg whole body averaged).

MR imaging safety guidelines for patients with DBS systems who are undergoing imaging of the brain have been developed (26). This current in vitro study allows us to expound on some of the recommendations. The neurostimulation systems must be checked before imaging is performed and interrogated for proper performance of all components. There must be no broken electrode leads, connectors, or extensions. The amplitude for each neurostimulation system must be programmed to 0 volts, and the output turned to off before entering the magnetic field environment. Thus far, the published data about imaging examinations of the brain concerns the use of 1.5-T MR imaging systems only. The most thorough support for safety of various clinical MR imaging sequences is restricted to MR imaging systems with a head transmit-receive coil architecture. The safety of other MR imaging systems and other examinations in the body has not been established. With the head transmit-receive system, the local SAR should not exceed 2.4 W/kg (approximately 0.09 W/kg whole body averaged SAR) for any brain imaging sequences. Sequences should always be performed at the lowest possible SAR levels. It seems that most routine clinical imaging of the brain techniques may be safely performed within the above SAR restrictions, including fast spin-echo, spin-echo, gradient-echo, magnetization prepared rapid acquisition gradient-echo, echo-planar, diffusion, perfusion, and functional MR imaging. Quantitative magnetization transfer imaging techniques should be considered prohibited. MR angiographic techniques should not use magnetization transfer background suppression. Patients should be instructed to immediately report any unusual sensations that may occur during the MR imaging examination. Patients must be continuously monitored by visual and verbal means. Extreme caution is recommended concerning conscious sedation. After MR imaging is completed, the neurostimulation systems must be interrogated again to verify functionality and then reprogrammed to previous stimulation parameters.

Conclusion

In vitro testing suggests that many of the MR imaging sequences in our clinical imaging armamentarium can be performed safely in patients with DBS electrode implants by using head transmit-receive coils on 1.5-T MR imaging systems, with acceptable minor temperature elevations. There are complexities, variabilities, and uncertainties about predicting the MR imaging-related heating associated with quantitative magnetization transfer and magnetization transfer-suppressed MR angiography sequences, and it is recommended that magnetization transfer imaging techniques be avoided until more specific testing establishes safe parameters.