MR Imaging and Cochlear Implants with Retained Internal Magnets: Reducing Artifacts near Highly Inhomogeneous Magnetic Fields

SA-CME LEARNING OBJECTIVES

After completing this journal-based SA-CME activity, participants will be able to:

• ■ Discuss the risks of proceeding with MR imaging for patients with a cochlear implant.

• ■ Identify common MR imaging artifacts arising from cochlear implants with typical neuroimaging sequences.

• ■ Describe strategies to reduce or shift artifacts from an anatomic region of interest.

Introduction

Cochlear implants and auditory brainstem implants (ABIs) have dramatically improved the quality of life for many patients with severe to profound sensorineural hearing loss. When these patients require a magnetic resonance (MR) imaging examination, the implanted components of the device introduce complexity to both patient preparation and image acquisition. The internal components of the implant—the active circuitry, electrodes, receiver antenna, and magnet—interact with various components of the MR imaging unit. Safety and image quality concerns exist when performing an MR imaging examination on a patient with unilateral or bilateral cochlear implants (1,2).

Many early cochlear implant models have never been MR imaging safety tested and are still considered MR Unsafe (using the ASTM International standard) in any MR imaging environment (3). However, several manufacturers now offer cochlear implants containing an internal magnet that have both U.S. Food and Drug Administration (FDA) approval and European Economic Area (EAA) CE marking as ASTM MR Conditional with specific conditions to image at 1.5 T, and at least one device is also MR Conditional for 3 T. Cochlear implants that contain a small magnet create a particular challenge at MR imaging. The patient may experience significant discomfort due to movement of the implant magnet and may need additional preparation before the examination. The local magnetic field from the implant also makes image artifacts notably worse. Signal loss can extend more than 10 cm from the implant when using a spin echo–based sequence, an imaging technique that is typically robust in the presence of magnetic field inhomogeneities (4). Artifacts from more-advanced imaging techniques, such as parallel imaging or long echo train imaging, may be subtle and located in regions away from the primary artifact.

Evaluation of tumors near the internal auditory canal and cerebellopontine angle requires fat-saturated imaging techniques, balanced gradient-echo acquisitions, and, potentially, diffusion-weighted imaging. All of these imaging techniques are highly sensitive to magnetic field inhomogeneities, and the implanted magnet often produces profound artifacts that may obscure adjacent regions of anatomy.

In this article, we describe the interaction of cochlear implants with the MR imaging environment. We also describe patient populations with cochlear implants that require routine MR imaging, how to prepare the patient to enter the MR imaging unit, and strategies to best perform a diagnostic MR imaging examination. Phantom studies demonstrate artifacts common to cochlear implants and are presented with corresponding clinical images to help illustrate artifact reduction strategies. MR imaging of patients with retained cochlear implant magnets is technically demanding, but with close collaboration among the radiologist, MR imaging technologist, and medical physicist or MR imaging scientist, the examination can be carefully tuned to the patient’s needs.

Cochlear Implants and the MR Imaging Environment

Cochlear implants are designed to bypass defective inner ear hair cells, sending electrical signals directly to the terminal ends of the cochlear nerve located within the cochlear modiolus. ABIs are of similar design but instead transmit signals directly to the cochlear nucleus in the brainstem and are most commonly used in patients with an absent cochlear nerve. As of 2012, more than 300 000 patients worldwide have received cochlear implants for rehabilitation of severe to profound sensorineural hearing loss (1). A complete cochlear implant or ABI system consists of both internal and external components. The external speech processor detects sound and transmits signals to the internal device transcutaneously. The internal device contains a radiofrequency (RF) antenna to receive signals, a magnet to secure the external speech processor in the proper position, an electronics package to transmit the processed signal, and an electrode to convey the signal to the cochlea or brainstem (Fig 1).

Figure 1a.

Download as PowerPoint

Figure 1b.

Download as PowerPoint

Figure 1c.

Download as PowerPoint

Historically, the internal magnet had to be routinely removed before performing MR imaging and replaced after the examination was complete. This strategy reduces associated artifacts and discomfort from magnet movement. However, drawbacks include the need for two minor surgical procedures (magnet removal and replacement), an increased risk of infection, and a period of nonuse while the incision site heals. More recently, many centers have begun performing MR imaging with a retained internal magnet.

To reduce the risk for magnet displacement when the internal magnet is left in place, a head wrap or splint is often used during MR imaging (5). An internal magnet system that allows the internal magnet to freely rotate and align with the constant magnetic induction field (B₀) has also been developed (Synchrony; MED-EL, Innsbruck, Austria), thereby reducing torque effects. Even with such precautions, patients may experience significant discomfort due to the movement of the magnet in the MR imaging environment. Some implant manufacturers make recommendations regarding the suitability of proceeding with MR imaging on the basis of the bone thickness adjacent to the magnet. When considering MR imaging for pediatric patients with a cochlear implant or ABI, tolerance of a tight head wrap, potential for discomfort, and bone thickness may be part of the decision of whether to proceed with MR imaging.

The implant manufacturer’s MR imaging conditions for any particular implant model are highly specific to that model and may intermittently be updated to reflect new testing or experience. The best clinical practice is to verify the implant model and current MR Conditional guidance each time a patient with a cochlear implant presents for MR imaging.

The three main subsystems of the MR imaging unit, the static magnetic field (B₀), the RF field, and changing gradient magnetic fields, may interact with the internal components of the implant, and efforts can be made to minimize negative effects. These MR imaging subsystems and their associated safety precautions are described in depth by the International Electrotechnical Commission standard IEC 60601-2-33, edition 3.2 (3). This standard also mandates what safety information the MR imager manufacturer must provide in the instructions for use, including data for determining whether the particular imaging unit can accommodate the MR imaging conditions specified by the implanted device.

The B₀ may cause the patient to experience a pulling sensation at the site of the magnet and torque, perceived as pressure, as the magnet rotates to align with the strong B₀, both of which may subject the patient to considerable local discomfort. If the imaging unit table is of dockable design, patients with cochlear implants may be positioned for the examination on the table outside of the MR imaging room, away from the fringe field of the magnet. The fringe field of the MR imaging unit within 1 meter of the mouth of the bore produces the strongest translational force on the internal magnet. The technologist should avoid repositioning the patient near the mouth of the imaging unit bore, where the static magnetic field has the largest spatial gradient and causes the most discomfort due to force and torque imparted on the implanted magnet.

Once the patient is advanced into the bore, RF energy during active imaging may cause localized heating within the electrode and conductive loops of the antenna. Limiting RF deposition according to the implant manufacturer’s recommendations can mitigate these effects. The changing gradient magnetic field could induce currents within the implant circuitry; as active electronic devices, these effects may lead to the patient perceiving clicks and odd noises during image acquisition.

Despite the precautions specified in the MR Conditional labeling, surgical revision occasionally is required after completion of the MR imaging examination to correct magnet displacement (1,7,8). If the internal magnet has been removed and replaced multiple times, the silicone flange that secures the magnet may become compromised or fatigued, and the magnet may be more likely to displace or flip from its designated position when approaching the static magnetic field of the MR imaging unit.

Some degree of patient discomfort during MR imaging is not uncommon, and, given the risk that surgical revision may occasionally be needed after the examination, a risk-benefit analysis should be performed before each MR imaging examination for patients with a cochlear implant or ABI. If alternative diagnostic modalities such as computed tomography (CT) or ultrasonography may provide answers to the clinical question, these could be considered before attempting MR imaging.

Hearing Loss–related Diseases Best Monitored with MR Imaging

Hearing loss may stem from progressive disease processes that require frequent MR imaging surveillance and are often close to the cochlear implant or ABI device. Familiarity with disease manifestations, MR imaging techniques used to evaluate disease progression, and artifacts associated with these imaging techniques can help the team prepare an efficient MR imaging examination for a patient who may be very uncomfortable for the duration of the examination. Patients with neurofibromatosis type 2 (NF2), an inherited autosomal-dominant syndrome characterized by multiple schwannomas, meningiomas, and ependymomas, are often candidates for a cochlear implant or ABI. The hallmark of NF2 is bilateral vestibular schwannomas, which commonly result in progressive sensorineural hearing loss. As NF2-associated tumors also occur at other intracranial, skull base, and spinal locations, these anatomic regions also generally require frequent MR imaging surveillance (Fig 2).

Because of the proximity of the cochlear implant or ABI to the posterior fossa, patients with NF2 pose special challenges in performing diagnostic MR imaging, owing to artifacts and image degradation produced by the retained magnet (9). The artifacts may completely obscure the anatomy most in need of monitoring. These patients typically require at least yearly MR imaging examinations to evaluate for possible disease progression or treatment effects.

Figure 2a.

Download as PowerPoint

Figure 2b.

Download as PowerPoint

Figure 2c.

Download as PowerPoint

Patient Preparation for MR Imaging

Patients should be instructed to keep their reading glasses (if applicable) and the cochlear implant or ABI external speech processor with them to aid communication until they are ready to enter zone IV (the MR imaging room), as they may be functionally deaf without the external speech processor, which is labeled as MR Unsafe for all current models (10).

The MR imaging technologist should provide additional time before the examination to coach the patient on the expected sensations associated with the implant and to apply a splint with the assistance of a member of the otolaryngology team, if required. Depending on the required splint and head wrap design, significant pressure at the site of the internal magnet may be uncomfortable for the patient. The splint should be applied as a last step before the patient enters the imaging room to maximize the available imaging time. If directed by a physician, a local anesthetic may provide some relief for discomfort from the splint. A pneumatic call system is often used within the MR imaging unit to enable communication between the patient and the technologist. We recommend demonstrating and testing the call system once the patient is in position. Testing the call button serves three purposes: it confirms that the call system is functional, that the patient understands how to use it, and that the patient is physically capable of using it.

After completion of the examination, be prepared to immediately remove the splint. The external speech processor can be used as a basic test of functionality and to assess whether the cochlear implant magnet has been displaced. Palpation of the scalp at the site of the implant may also provide an indication of magnet position. In rare cases, radiography may be helpful for answering questions of whether the magnet remains in the proper location within the silicone sleeve (7). Pain at the implant site after the examination has been completed is often the first sign of a displaced magnet. If a patient reports pain or discomfort when using the device, this should be carefully investigated to ensure that the magnet is seated properly.

Typical MR Imaging Protocols and Associated Artifacts

With the internal magnet of the cochlear implant or ABI degrading the homogeneous B₀ field necessary for MR imaging signal localization, significant artifacts should be anticipated even with the most robust acquisitions, including spin-echo–based techniques. Figure 3 shows typical acquisitions from a routine brain protocol on a 1.5-T MR imager and the artifacts generated from the presence of a cochlear implant with an internal magnet. The highly nonlinear magnetic field imposed by the internal magnet is beyond the correction capabilities of the linear gradient shim fields available from any MR imaging unit manufacturer. Several artifacts from the inhomogeneous B₀ field are evident: intravoxel dephasing, where protons within a single voxel precess at different frequencies, resulting in signal loss; geometric distortion of the section selection pulse; and in-plane distortion for lower-bandwidth acquisitions, causing signal pile up or hot spots at some locations and stretched-out low signal in other locations (11). Distortion of the location of section selection is most evident for this examination on the sagittal T1-weighted spin-echo image (Fig 3a), where the magnetic field on the superior side of the internal cochlear implant magnet adds to B₀, shifting the section selection to tissue medial to the intended section, and the magnetic field inferior to the cochlear implant magnet subtracts from B₀, shifting the section selection lateral to the intended section (12,13). The final image displays a gross discontinuity where nonadjacent tissue is reconstructed on the image near the location of the cochlear implant magnet.

Figure 3a.

Download as PowerPoint

Figure 3b.

Download as PowerPoint

Figure 3c.

Download as PowerPoint

Figure 3d.

Download as PowerPoint

Figure 3e.

Download as PowerPoint

Figure 3f.

Download as PowerPoint

MR imaging datasets acquired with cochlear implants or ABIs present may not be suitable for precision surgical planning, such as might be needed for stereotactic radiosurgery or proton-beam surgical procedures. Even without the internal magnet, moderate metallic artifact from the implant antenna should be anticipated. Owing to the anticipated artifact, experienced implant surgeons frequently implant new cochlear implant or ABI systems as far away as possible from intracranial pathologic conditions that may require ongoing MR imaging surveillance or treatment. For example, implanting the device farther posterior and superior may allow better long-term monitoring of the internal auditory canal (14).

Recommended MR Imaging Protocol Modifications for Cochlear Implants and ABIs

Fat Suppression

Fat-saturation techniques rely on the chemical shift of the proton resonance of fat relative to water. A chemical shift–selective (CHESS) RF pulse tuned to fat (220-Hz frequency shift at 1.5 T) saturates and dephases the fat signal (15). The CHESS pulse depends on a very homogeneous B₀ field; any tissue that is 220-Hz off resonance is also saturated. Even in the absence of an implant, anatomically induced B₀ offsets due to anatomic shape or differing magnetic susceptibilities can be challenging for CHESS pulses, in some cases causing the CHESS pulse to instead suppress water signal, or shifting the pulse off resonance in the opposite direction such that it no longer suppresses fat or water. Figure 4 shows a map of the magnetic field within the phantom, where the signal intensity swaps from white to black at 220-Hz intervals. A fat-saturation sequence demonstrates signal loss corresponding to the B₀ off resonance at the same locations within the phantom as the signal intensity swaps. Likewise, in vivo, we observe regions of suppressed water signal and bright fat signal throughout an image that ideally would have uniformly dark fat (Fig 4c, 4d).

Figure 4a.

Download as PowerPoint

Figure 4b.

Download as PowerPoint

Figure 4c.

Download as PowerPoint

Figure 4d.

Download as PowerPoint

When imaging near the cochlear implant or ABI magnet, fat-suppression techniques such as short tau inversion time inversion-recovery (STIR) or robust Dixon techniques should be substituted for fat-saturation techniques.

STIR relies on the short T1 relaxation time of fat, using an inversion-recovery sequence to measure data when fat is dark (16). Dixon techniques collect data at multiple TEs to resolve signal originating from the 220-Hz fat-water chemical shift (17–19). The Dixon reconstruction algorithm for fat- and water-only images may also struggle with severe B₀ inhomogeneity, but if both fat and water datasets are saved, any spurious fat/water signal swaps could be sorted through by the interpreting radiologist. Figure 5 shows Dixon images acquired in the same patient as in Figure 4 at a subsequent MR imaging examination. Hybrid fat-suppression techniques that rely on both fat saturation and the short T1 signal properties of fat may more reliably suppress fat signal but will still saturate water signal at some locations. Although fat saturation with a CHESS pulse is commonly used for fat suppression within the head and neck, CHESS is the fat-suppression technique most sensitive to magnetic field inhomogeneity.

Figure 5a.

Download as PowerPoint

Figure 5b.

Download as PowerPoint

Figure 5c.

Download as PowerPoint

CISS or FIESTA-C

Balanced steady-state free precession–based imaging techniques have band artifacts that appear at off-resonance frequencies that depend on the TR. A shorter TR spreads the band artifacts farther apart. Even with phase-cycling schemes, such as constructive interference steady state (CISS) (Siemens Healthineers; Erlangen, Germany) or fast imaging employing steady-state acquisition with cycled phases (FIESTA-C) (GE Healthcare), designed to reduce signal loss within the band artifacts, residual bands remain near the large off-resonance band created by the implant magnet (20,21). The band artifacts may interfere with visualization of the anatomy of interest. As these imaging techniques are workhorse sequences for evaluation of the internal auditory canal, such artifacts are problematic. Figure 6 demonstrates band artifacts in a phantom and in vivo.

Figure 6a.

Download as PowerPoint

Figure 6b.

Download as PowerPoint

Figure 6c.

Download as PowerPoint

A few strategies to shift the artifact away from the anatomy of interest include adjusting imaging parameters to change the TR, adjusting the shim using a local shim volume during tuning, and manually overriding the applied shim. When manually adjusting the shim, depending on the implant location, often a simple doubling of the applied shim in the left-to-right direction is adequate to move the artifact.

Parallel or Accelerated Imaging Artifacts

Parallel or accelerated imaging relies on the location of the receiver coil elements to help spatially encode the image and is an integral part of many modern MR imaging sequences (22). Parallel imaging reconstruction uses calibration scans to map the relative signal detected by each coil element. The calibration scan may be acquired separately or integrated into the sequence itself. Typically, the calibration scan is a low-resolution gradient-echo acquisition that suffers from intravoxel dephasing. With signal dropout near the implant magnet, the parallel imaging reconstruction fails to spatially encode signal in that region, creating aliased signal on the reconstructed image at predictable locations based on the acceleration factor. Figure 7 shows a phantom image acquired with and without parallel imaging and an in vivo image demonstrating the parallel imaging artifacts. With an acceleration factor of 2, the aliased signal is shifted in the phase-encoding direction to a location half of the field of view from where the signal originated. Depending on the imaging parameters, techniques such as generalized autocalibrating partially parallel acquisition (GRAPPA) with integrated oversampling instead of a calibration scan may have fewer artifacts (23).

If possible, avoid accelerated acquisitions when imaging a patient with a cochlear implant or ABI, as receiver coil elements near the internal magnet do not contribute to the spatial encoding as might otherwise be expected.

Figure 7a.

Download as PowerPoint

Figure 7b.

Download as PowerPoint

Figure 7c.

Download as PowerPoint

Figure 7d.

Download as PowerPoint

Metal Artifact–reduction Techniques

Metal artifact–reduction techniques such as multi acquisition with variable resonance image combination (MAVRIC) are designed to handle B₀ inhomogeneity near large metallic implants of varying magnetic susceptibility (24–26). Typically, these techniques acquire data at several frequency offsets to map the off-resonance data back onto the primary image. Variations of the technique may use view-angle tilting to also correct the distorted section selection. These techniques do require additional acquisition time to capture the frequency offsets and are tuned to the frequency offsets of common orthopedic implants, which are not a large enough range to resolve the B₀ inhomogeneity induced by the implant magnet. However, the size of the artifact is markedly reduced and may allow evaluation of anatomic structures that would otherwise be obscured by artifact. In comparison, three-dimensional T1-weighted fast spin-echo (FSE) sequences using variable flip-angle techniques are highly sensitive to off resonance as the refocusing pulses get out of phase (Fig 8) (27).

Figure 8a.

Download as PowerPoint

Figure 8b.

Download as PowerPoint

Figure 8c.

Download as PowerPoint

Figure 8d.

Download as PowerPoint

Figure 8e.

Download as PowerPoint

Figure 8f.

Download as PowerPoint

Conventional wisdom suggests increasing the receiver bandwidth to resolve structures near metal or other B₀ inhomogeneities. However, the highly nonlinear B₀ field near the small cochlear implant or ABI magnet changes rapidly, so there is only a modest benefit of about a centimeter to increasing the receiver bandwidth, at the cost of reduced signal-to-noise ratio across the image. If the operator is not careful about how the receiver bandwidth changes other imaging parameters, the operator may be surprised with the resulting image contrast. For example, the echo train length for an FSE sequence may become too short to obtain the desired effective TE, providing more of a proton-density–weighted appearance than a T2-weighted appearance. Figure 9 shows the FSE images of the phantom with receiver bandwidths of 16 kHz and 50 kHz. The echo train length after each 90° excitation pulse had to be doubled to obtain the desired TE. With only slight benefit near the magnet, but overall image-quality reduction, increasing the receiver bandwidth adds little value and risks unforeseen drawbacks across the rest of the image. Increasing the receiver bandwidth should generally only be used as a problem-solving tool or to reduce signal pile up to evaluate tissue near the implant.

Figure 9a.

Download as PowerPoint

Figure 9b.

Download as PowerPoint

Echo-planar and Diffusion-weighted Sequences

Echo-planar imaging is notoriously sensitive to B₀ off resonance. With its rastered acquisition collecting multiple lines of k-space following a single RF excitation, the phase-encoding direction is essentially doing a very-low-bandwidth acquisition (28). This manifests as significant geometric distortion and intravoxel dephasing adjacent to normal anatomic structures of differing susceptibility. In the presence of a cochlear implant or ABI, the diffusion-weighted echo-planar images are often rendered nearly uninterpretable, with only small regions of clear anatomy visible through the smeared mess. However, given the brief acquisition time and the unique ability to depict early infarction, most radiologists would still recommend attempting diffusion-weighted echo-planar imaging.

Other data acquisition schemes, such as periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER; GE Healthcare) or BLADE (Siemens Healthineers), acquire the data in multiple shots, traversing k-space in a unique fashion (29,30). The pattern of artifact is quite different from that produced at echo-planar imaging and may prove useful should the patient need assessment for stroke. Figure 10 shows diffusion-weighted images acquired with echo-planar imaging and with PROPELLER in a patient with a right-side cochlear implant.

Figure 10a.

Download as PowerPoint

Figure 10b.

Download as PowerPoint

Figure 10c.

Download as PowerPoint

Figure 10d.

Download as PowerPoint

Conclusion

Cochlear implants or ABIs with retained internal magnets present significant MR imaging challenges, particularly when the disease to be imaged is close to the implant. Extreme care must be used to position the patient for the MR imaging examination and to secure the implant when an internal magnet is included. The implant manufacturer’s MR imaging conditions for each individual implant model should be carefully followed, as subtle nuances in implant design may affect the likelihood of the internal magnet migrating or the components heating during the examination. Depending on the implant location and orientation, the patient’s perception of pain near the implant, and the need for a splint or head wrap, each patient has her or his own tolerance level for remaining in the imager. Not all patients with cochlear implants can tolerate the sensation of pressure at the site of their implant, and some may choose to abort the examination.

Moderate to severe artifacts are to be expected, both near the small internal magnet and displaced away from the magnet, depending on the image acquisition technique. Even with the help of a dedicated MR imaging physicist or scientist, the artifact will never be completely resolved. During surgical planning, experienced cochlear implant surgeons adjust the planned location of the implant with the hope of limiting artifact near disease that requires ongoing medical attention.

When imaging the head or neck near an implant with a retained magnet, STIR or Dixon techniques provide more-reliable fat suppression than CHESS fat-saturation pulses. If possible, parallel or accelerated imaging techniques that rely on coil sensitivities for spatial encoding should be avoided. Metal artifact–reduction techniques may help reduce the artifact but are unlikely to fully compensate for a magnetic material. Consider acquisitions other than echo-planar imaging to avoid gross geometric distortion and signal loss. Most importantly, when adjusting imaging parameters to reduce artifacts, always carefully check that imaging is still within the implant manufacturer’s MR imaging conditions for RF deposition, and avoid parameter changes that substantially alter the desired image contrast. Being prepared with strategies to reduce or move the B₀ inhomogeneity artifact can shorten the duration of the examination and allow diagnostic images to be obtained.