Cardiovascular magnetic resonance guided electrophysiology studies

CMR has been used most extensively to assist planning and guidance of atrial fibrillation (AF) ablation procedures. AF is the most common clinically relevant arrhythmia affecting 0.4% of the general population [6]. The principle morbidities related to AF are stroke due to embolization of atrial thrombus and symptoms related to poor heart rate regulation with resting heart rates commonly over 110 beats per minute. In the early 1990's surgical modification of the atria with a series of linear incisions was found to be effective at controlling AF, but a minimally invasive catheter based procedure could not replicate these results [7, 8]. It was later recognized that the triggering foci for AF frequently arise from one or more pulmonary veins (PVs) [9]. The ability to cure AF by ablating PV triggers or ablating conduction pathways exiting the PV's was promising but hampered by the risk of pulmonary vein stenosis due to injury of the vessels [8]. Electrospatial mapping technology led to the development of purely anatomic circumferential ablation strategies in which circular lesions are created further from the PV ostia to block the exit of PV triggers [4] (Figure 1a). Using this technique alone or in combination with PV isolation, a 70% to 80% success rate has been achieved [8]. However, repeat procedures are often needed to achieve this success and the success rate drops to 50% or less for the more chronic forms AF associated with ischemic, hypertensive, and valvular heart disease [8]. There also remains a 5% risk of major complications including cardiac perforation, pulmonary vein stenosis, and the rare but potentially lethal risk of atrioesophageal fistula formation [8].

In an effort to improve procedural success and reduce complications, 3-D CMR angiography (MRA) has been used to assist planning of AF ablation. Kato and colleagues used MRA to study left atrial anatomy in normal subjects and patients with paroxysmal atrial fibrillation and found that 38% of people had pulmonary vein anatomic variants [10]. Identification of these variants is important because AF triggering foci can be located within additional veins (Figure 2a). In addition, ablating near small or early branching PVs increases the risk of pulmonary vein stenosis [11]. Identification of certain anatomic variants before the procedure can also assist in catheter selection or favor using the circumferential ablation approach which is less affected by variant anatomy. 3-D imaging may also reduce the risk for complications by visualizing the relationship of the left atrium to surrounding structures including the esophagus, descending aorta, right pulmonary artery, and left circumflex coronary artery [12‐15]. Knowing the location of these structures can be used to direct placement of ablation lesions to lower risk areas or guide reduction of ablation power when lesions are place close to these structures.

CMR also has the potential to guide the treatment of scar based monomorphic ventricular tachycardia (MVT), a potentially lethal arrhythmia that is difficult to treat medically or with current ablation techniques. Ventricular tachycardia that results in uniform repetitive electrical activation of the heart arises from anatomically fixed arrhythmia substrate that can be targeted for ablation. Myocardial scarring due to infarction, cardiomyopathy, sarcoidosis, ARVD, or cardiac surgery is a common cause of MVT [16]. Scar related MVT typically depends on critical isthmuses of conductive tissue bounded by non-excitable scar or a valve annulus [17]. Ablating isthmus pathways can be curative but identifying the pathways using traditional mapping techniques can be difficult because these arrhythmias often lead to hemodynamic collapse. Substrate based approaches utilizing electrospatial mapping to identify reduced voltage scar border zone areas and isolated diastolic potentials within low voltage scar are now being used to identify critical portions of the arrhythmia circuit to target ablation [5, 16] (Figure 1b). Still, ablation of MVT can be arduous. In addition to requiring careful point-by-point electrical mapping of the endocardium, rhythms resulting from epicardial pathways may require additional epicardial mapping and rhythms resulting from intramural pathways may be inaccessible to electrical mapping. In addition, procedures commonly last over six hours to achieve cure rates on the order of 70% even in the most experienced hands and success rates can be considerably less in lower volume centers.

The use of CMR for assisting MVT ablation is still in the investigational stages but shows promise. Late gadolinium enhancement CMR (LGE-CMR) has been used extensively to characterize regions of scar in ischemic and non-ischemic cardiomyopathy (Figure 2b). A number of clinical studies have demonstrated the association of LGE-CMR scar characteristics such as size, transmurality, and border-zone area with the risk of MVT [18‐20]. Recent work suggests that high resolution LGE-CMR can be used to more directly assist in MVT ablation planning. Ashikaga and colleagues used an epicardial sock with around 300 electrodes to obtain high-resolution electrical maps during MVT in a pig infarct model [21]. These maps were registered to very-high resolution (0.39 × 0.39 × 0.39 mm) ex-vivo LGE-CMR to assess the relationship of MVT electrical propagation to scar morphology. Detailed scar imaging revealed numerous previously unseen features including 3-D tracts of viable myocardium within scar and scar within viable myocardium that visually correlated with the MVT isthmus identified by electrical activation mapping. Ciaccio et.al. complimented these findings by computationally predicting suitable locations for MVT ablation with the use of high resolution LGE-CMR scar imaging [22]. Using a model that incorporates regional scar thickness to estimate MVT excitation wavefront propagation, the MVT circuit isthmus was predicted and shown to overlap the actual isthmus, observed by electrical mapping, by around 90%. Though these experimental studies use significantly higher resolution LGE-CMR maps than the typical 1.5 × 2.5 × 8 mm pixel resolution used clinically, methods to obtain higher resolution scar images in patients are being developed and will be discussed further below [23, 24]. Together with studies to identify safe MR imaging procedures for patients with implanted defibrillators [25] and clinical correlation of LGE-CMR scar morphology to successful VT ablation sites, the role of CMR for clinical VT ablation should be better defined in the near future.

The detailed anatomic information available from CMR is now being used as a roadmap for guiding placement of ablation lesions [26‐30]. A number of techniques have been developed to register electrospatial mapping catheter coordinates to pre-acquired 3-D CMR and CT images. Dong and colleagues recently reported their technique in patients undergoing AF ablation [30]. They felt that 3-D imaging was helpful for tailoring ablations to the variant PV anatomy found in 47% of patients. They also noted that 3-D images of the atria helped to guide lesion placement in areas where stable catheter positioning was difficult, such as along the tissue ridge separating the left atrial appendage from the left pulmonary veins [11, 30]. However, even when guided by 3-D image roadmaps, the study noted that circumferential ablation around the PVs prevented escape of PV triggers in only 32% of patients [30]. To completely isolate PV triggers, additional electrical mapping and ablation of specific conduction pathways was required. Others have created more extensive, riskier lesions using saline irrigated ablation catheters and still report PV isolation rates of only 50 to 60% after circumferential ablation around the PVs [31, 32].

These sub-optimal results call attention to limitations of this current state-of-the-art in 3D image guided cardiac ablation. First, pre-acquired roadmap images may not correspond to the anatomy during the procedure. Changes in cardiac chamber size associated with variations in heart rate, rhythm, and volume status are not accounted for by pre-acquired imaging and could lead to catheter position registration errors [29]. Additional registration errors can result from patient motion during the exam, respiratory motion, and beat-to-beat motion of the heart, including significant motion of the PVs [33]. Second, marking attempted ablation positions and confirming reduction in the local electrogram voltage does not necessarily establish creation of a permanent ablation lesion or a continuous ablation line [30]. The electrode tissue contact area and electrode exposure to flowing blood are important factors in forming adequate ablation lesions [34] but are poorly assessed by fluoroscopy and electrospatial mapping guided procedures. Ablation lesion extent, unintentional gaps in ablation lines, and transient lesion components such as edema similarly are not well predicted by current techniques, including intra-cardiac ultrasound. These factors limit ablation accuracy and have been shown to reduce procedure efficacy [35, 36].

Intracardiac echocardiography (ICE) addresses some of these shortcomings and is increasingly used in clinical practice [37]. ICE has been used to visualize electrode tissue contact, an important factor for efficient ablation lesion creation. Visualization of microbubbles at the electrode tissue contact interface during ablation has been also been used to indicate adequate electrode tissue contact while the presence of more coarse bubbles has been associated with inappropriately high tissue temperatures that could lead to tissue charring and coagulum formation. However, ICE has limitations for guiding ablation. ICE requires invasive placement and manipulation of a separate imaging catheter and physical limitations on image plane orientation and field of view limit its ability to evaluate lesion extent and characterize extended regions of ablation. Also the ability of ICE to reliably distinguish regions of ablation from surrounding viable tissue has not been established.

Perhaps the most significant advantage of CMR-guided ablation therapy is the potential to visualize ablation lesions with high spatial and temporal resolution. The typical endpoint of current ablation procedures is absence of electrical conduction across the ablated region and/or an inability to re-induce the clinical arrhythmia with cardiac pacing and medications. However, propagation of electrical signals through the heart is affected by a number of factors including the tissue temperature change induced by ablation [43, 44]. Some of these factors may be reversible over time leading to arrhythmia recurrence [35, 36]. As described below, CMR appears capable of delineating areas permanent tissue damage caused by ablation. Using CMR lesion imaging to guide ablation could improve the procedure endpoint from assessment of potentially transient electrophysiologic changes to a more direct assessment of complete lines of permanently damaged tissue in the region of interest.

500 kHz radiofrequency (RF) current is the most commonly used ablation source used for electrophysiology procedures. Cryothermy, ultrasound, laser, and microwave ablation are also being investigated. Ablation lesions can be visualized because CMR is able to detect specific changes in proton precession and relaxation properties resulting from heating and heat induced biophysical changes in cardiac tissue including interstitial edema, hyperemia, protein conformational changes, cellular shrinkage, and tissue coagulation [38]. Acute interstitial edema is likely responsible for the hyperintense region corresponding to the area of acute RF ablation damage observed by T2-weighted FSE imaging [38, 45] (Figure 5). Dickfield and colleagues found that this hyperintense region correlated well with necrotic lesion size on gross pathology and also noted that gaps between lesions on imaging corresponded with lesion gaps on pathology [46]. Lesion visualization by T2 weighted imaging has been reported as soon as 2 minutes after ablation and stable imaging characteristics have been observed from 30 minutes to 12 hours post ablation [38, 46]. This could make T2 weighted CMR a tool to evaluate lesions and lesion continuity over the course of an ablation procedure.

T1 weighted non-contrast enhanced MR imaging of RF ablation lesions has also been investigated [46]. T1 changes may be related to tissue heating, protein denaturation, and changes in intracellular and extracellular free water distribution during and following ablation [45‐47]. Similar stability of imaging characteristics were reported from 30 minutes to 12 hours following ablation, though the lesion contrast by T1 weighted imaging appears to be less than for T2 weighted imaging [46].

LGE-CMR can provide better visualization of RF ablation lesions compared with non-contrast imaging techniques (Figure 6). The time to achieve full enhancement of RF ablation lesions, one to two hours, is considerably longer than for DECMR of myocardial infarct scar [48]. However, good correlation with pathologic lesion size was noted for intermediate enhancement patterns from one minute to 2 hours after contrast injection, allowing lesion extent to be assessed without waiting for full enhancement [48]. The 1 to 2 hour interval required for renal clearance between repeated dosing of gadolinium and the ceiling on total allowable gadolinium dose limits the use of this technique for serial lesion assessment during a procedure [45]. Still, gadolinium enhanced imaging may be useful for evaluating gaps in ablation lines after completion of a procedure to assess the need to place additional lesions.

Other methods for monitoring ablation lesion formation during RF energy application are also being investigated. Proton resonance shift thermography is an CMR technique that takes advantage of the decrease in the proton resonance frequency with increasing temperature [49]. This technique has been used to follow tumor ablation in the uterus, liver, prostate, and brain using diverse energy sources including RF, high frequency ultrasound, laser, and microwave [50‐55]. Its use for following RF ablation in the beating heart is being investigated. Current-vector mapping has also been described for monitoring the extent of tissue power deposition during RF ablation [56].

While most cardiac ablation lesion CMR studies have been performed in roughly 10 mm thick ventricle, imaging the less than 3 mm thick human atria is of particular clinical interest given the difficulty of achieving long term pulmonary vein isolation following atrial fibrillation ablation. Peters et.al. demonstrated 3D LGE-CMR of left atrial ablation lesions 10 to 15 minutes after contrast injection using image based respiratory gating [24]. This gating technique, also known as respiratory navigator imaging, allowed higher resolution 3D imaging to be performed without the need for prolonged breath-holding by tracking diaphragm position on fast 1-D images and collecting 3-D image data within a narrow range of diaphragm positions. Current applications have used a roughly 100 ms mid diastolic acquisition window timed to precede atrial systole to reduce atrial motion during imaging. Image resolutions of 1.25 × 1.25 × 2.5 mm (reconstructed to 0.6 × 0.6 × 1.25 mm) have been demonstrated. McGann, et.al combined this technique with 3-D visualization and quantification to assess left atrial scarring before and following atrial fibrillation ablation. They found that subjects with more than 13% left atrial delayed enhancement following ablation had a nearly 20 times higher chance of being free from atrial fibrillation than those with low amounts of post ablation delayed enhancement [57]. The technique is now being used to image immediately following the procedure to target further ablation and potentially reduce the need for repeat procedures. Interestingly, high resolution 3D LGE-CMR may also be useful for better identifying patients who will respond poorly to current ablation techniques [58]. Of note, such high-resolution imaging is not yet feasible in all patients. In the above studies, imaging took 5 to 10 minutes and 10% to 30% of patients were excluded from analysis because of poor image quality that was attributed to patient motion, significant arrhythmia, or incorrect inversion time selection. Techniques to improve the speed and reliability of high-resolution CMR require further investigation.

While cardiac gating can be used to generate MR images with excellent spatial resolution by splitting data collection over multiple heartbeats, real-time CMR requires a more deliberate trade off between temporal and spatial resolution. To adequately visualize catheters, CMR guided EP procedures require an in-plane spatial resolution of around 2 mm². The target temporal resolution is 7 frames per second (fps), the usual x-ray fluoroscopy frame rate for clinical EP procedures.

Since the initial 1 fps imaging used to guide the first MR guided EP procedure [38] faster, stronger gradients have increased the temporal resolution capabilities of fast gradient recalled echo sequences to the 5 fps range [59]. Improved gradient performance and B0 field homogeneity have also allowed real-time imaging to be performed with coherent steady state pulse sequences (ie. SSFP, true-FISP, or FIESTA). These sequences provide increased contrast to noise performance at a given frame rate compared with fast gradient recalled echo sequences [59]. Parallel imaging techniques can provide additional improvement in temporal resolution without sacrificing spatial resolution. These techniques accelerate imaging by covering the region of interest with multiple receive coils and using the different spatial sensitivities of these coils to correct for undersampling of image data [60]. The acceleration achieved from acquiring less data is countered by the increased processing time required for estimating coil sensitivities and for performing the parallel reconstruction [61]. Also, for a given number of receive coils, more undersampling leads to deteriorating image signal to noise or a reduced ability to suppress aliasing artifacts [62]. Balancing these issues, 15 fps true-FISP cardiac imaging with 128 phase encode lines can be performed using an 8 channel receive coil array and optimized reconstruction hardware [63]. Commercial CMR systems now commonly have multi-channel receivers and parallel imaging options. The performance of these systems is currently in the range of what is needed to perform CMR guided EP procedures at 5 fps with acceptable image quality [61].

While the current imaging rates are adequate for a single 2-D image plane, ideal visualization of the device, target anatomy, and surrounding reference anatomy may require multiple 2-D image planes or even 3-D imaging. Other techniques that can improve imaging speed while balancing imaging quality include non-Cartesian k-space sampling, temporal data sharing between images, and adjusting the tradeoff between temporal and spatial resolution [59]. These techniques may be particularly useful to accelerate imaging of reference anatomy views that are not depended on for device tracking. Use of 32 channel receive arrays to perform more rapid 3-D cardiac imaging and parallel transmission techniques to permit more efficient parallel data collection are also under active investigation [63‐65].

While fluoroscopy provides projection images where the entire catheter body and tip are easily visualized, 2-D MR images typically depict a slice through the body that is around 5 to 10 mm thick. Curved devices such as catheters may pass in and out of the MR imaging plane leading to misinterpretation of the device tip position. We have noted in preclinical studies that poor delineation of the tip position can result in tissue contact trauma, such as local hemorrhage. In addition, for electrophysiology ablation procedures the device tip contains the energy source. Misestimating the tip/tissue contact region can lead to inaccurate placement of ablation lesions.

During our feasibility studies, tip location has mostly been performed using interactive real-time sequences with a user interface that permits adjustment of the scan plan during image acquisition. Part of the catheter is first identified on some imaging plane and the plane is manually adjusted until the tip is located. For vascular procedures where the device is constrained to a co-planar segment of blood vessel, manual plane manipulation is acceptable since only minor image plane translations are needed to visualize the device tip and relevant anatomy. For navigation in cardiac chambers where the device tip location is less constrained, the frequent need for manual plane manipulation necessitates a skilled operator for image plane manipulation and can distract from efficient procedure workflow.

One approach to this problem is to automatically direct imaging to the device location using position sensors located in the catheter. 15 years ago Dumolin et.al. described using 1-D projection MR imaging along the x, y, and z directions to identify the 3-D position of small receiver coils located in a catheter tip [66]. This position tracking technique has since been interleaved within real-time imaging sequences to automatically move the image plane position to the catheter tip location during device manipulation [61, 67]. While a single tracking coil is sufficient to simply shift the image position, multiple tracking points are needed to maximize visualization of the device body or to orient imaging relative to the catheter tip direction. Multi-coil designs and tracking algorithms have been developed to reduce the need for separate matching circuits in space constrained catheter lumens [68, 69]. Other magnetic field and electric field based position finding techniques have been developed for medical device tracking that could also be used for catheter tracking in the CMR scanner [70‐72]. Some of these systems generate both position and orientation information for each sensor assembly [72, 73]. These systems may provide more accurate and catheter space efficient options for device tracking. They can also reduce the performance penalty and avoid the scanner specific complexities associated with interleaving tracking and real-time imaging sequences.

An alternative technique for device visualization uses non-slice selective imaging to produce an effect similar to projection x-ray fluoroscopy [74, 75]. The non-slice selective catheter imaging plane can be intersected with slice selective images containing the target anatomy to assist guidance of the catheter tip. This technique may be used to provide a "fluoroscopy" view of devices using a number of catheter antenna designs [69, 76‐78] (Figure 3a).

Another factor that affects device navigation in the CMR environment is the physical constraint of performing procedures near and within the narrow CMR bore. The availability of shorter, wider bore, high-field CMR scanners are making this less of an issue. Remote catheter steering is also gaining interest in the EP field to facilitate point-by-point electrical mapping of the cardiac chambers and assist stable device placement during ablation [79, 80]. Some of these techniques may be amenable to use in the CMR environment. A robotic catheter manipulation system that uses steerable sheaths with multiple pull wires was recently used to perform atrial fibrillation ablation in patients [81]. A magnetic remote steering technique has also been described that utilizes the torque generated by current carrying coils in the static CMR magnetic field to deflect a catheter tip [82, 83].

The ability to generate real-time images with arbitrary orientations in addition to anatomically detailed 3-D images with flexible tissue contrast makes CMR well suited for navigating complex arrhythmia anatomy and delineating complex ablation patterns. This flexibility also introduces the potential for disorientation and information overload. Appropriate displays and user-interfaces tailored to the workflow of an EP procedure are needed to manage this flexibility. Because thin-slice real-time imaging can intersect anatomy in unfamiliar ways, 3-D visualizations that plot real-time images oriented relative to reference images can be helpful (Figure 7). A basic imaging interface for MR-guided EP procedures would provide a convenient way to "bookmark" and access reference cardiac views, switch between real-time and lesion visualization sequences during the procedure, appropriately present lesion images for ablation line continuity assessment, and display the relationship of stored images, catheter position, and intracardiac electrograms characteristics in 3-D.

The most important consideration for any new diagnostic or therapeutic approach is safety. While a number of studies have been performed to determine the safety of conventional CMR with regard to electromagnetic energy exposure and tissue heating [84, 85], interventional procedures add additional considerations and raise new safety concerns [86‐89]. The most straightforward aspect to CMR device safety is the avoidance of ferromagnetic materials that could experience significant forces when brought close to the scanner. Though CMR unsafe objects, such as ferromagnetic scissors and needle drivers, may be needed during the preparatory phase of a procedure, until CMR compatible alternatives are available a system must be in place to methodically track and remove such objects before approaching the scanner. The electric current associated with defibrillation can also lead to strong displacement forces in high magnetic fields and should be performed with the defibrillator pads maintained a safe distance from the scanner bore [90]. Similar attention is needed to address the ferromagnetic properties and CMR electromagnetic interference compatibility of other equipment associated with electrophysiology procedures including physiology monitoring equipment, ablation and pacing sources, and anesthesia apparatus. Clear marking of high field areas and secure placement of objects that may experience magnetic forces is mandatory so that appropriate pieces of equipment are kept at a safe distance [91].

An additional safety concern particular to CMR guided cardiovascular procedures is the significant heating that can result from RF transmission induced current in extended metallic objects such as guidewires, wired electrodes, and metal braided catheters [87, 88]. This induction is more pronounced when portions of the device are located close to the RF transmit body coil housed within the edge of the scanner bore. Device length is an important factor in efficient coupling and heating; however, many other parameters can influence the sudden onset of significant heating in the setting of an interventional procedure [88, 92]. The simplest way to avoid this problem is to construct devices from non-metallic components when possible. Polymer materials for catheter braiding, such as Dacron and Kevlar, and composite materials for guidewires, such as glass fiber reinforced plastics, can be used to achieve device functional characteristics such as torqueability, stiffness, and kink resistance [38, 93]. Several approaches have also been developed to avoid significant induction heating in structures that require conductivity. Wires made from high resistance alloys and gold sputtered thread have been used to obtain intracardiac electrograms during imaging with significant reduction of electrode heating [94]. For structures where efficient power transfer is required, such as pacing or ablation electrodes, high frequency RF chokes can allow passage of signal lower than a few MHz while blocking unwanted MR transmit frequency currents [95, 96] A promising heating suppression technique for position tracking and intravascular imaging coils that need to pass differential mode signals at the same frequencies as unwanted common mode induced currents is to place thin transmission line transformers in the signal carrying cables [77, 97, 98]. Other strategies such as detuning and decoupling of circuits prone to heating [78, 99], fiberoptic transmission of signals [100], and use of inductively coupled resonators for wireless device tracking [101] are also considerations for new device design [25, 90]. Using surface coils instead of the higher power body coil for RF transmission has also been proposed as a way to reduce RF current induction in devices [102]. Now that academic sites and imaging companies are focusing on heating safe device development, more rapid progress in this area is expected.

Transitioning proof of concept studies to clinical electrophysiology procedures requires collaboration between academic centers, imaging and device companies, and regulatory agencies. The first CMR guided electrophysiology procedure in a patient was performed using custom catheters made to clinical specifications by a clinical catheter manufacturer [39]. Prior to use in a human an investigational device exemption was obtained from the FDA. As part of this device exemption, catheter heating and reduction of heating using RF filtering was demonstrated in a series of device positions and orientations within the scanner using high SAR imaging protocols [39]. In addition, safe use of the catheter in animals was documented prior to human studies. Though catheters with embedded imaging coils provided improved device visualization in the animal studies, these devices were not approved for or used in patients. Before generally applying CMR guidance to interventional electrophysiology, more standardized CMR compatible catheter safety guidelines and device testing protocols need to be developed.

The compatibility of CMR with implanted devices such as pacemakers and implantable cardiac defibrillators (ICDs) is also an important consideration in performing interventional CMR studies in the electrophysiology patient population. Particular concerns include static magnetic field induced movement of the device and scanning induced programming changes, device inhibition, activation of tachyarrhythmia therapies, and lead currents leading to heating or cardiac stimulation [103‐106]. Modern devices address some of these concerns with the use of less ferromagnetic material and improved resistance to electromagnetic interference [107]. A number of devices have been carefully studied during in-vitro and animal MR imaging and experience is growing for safe CMR in patients with selected devices under controlled scanning conditions [25, 107, 108]. This experience includes use of sequences relevant to modern CMR with SAR characteristics similar to those used for real-time CMR [25]. Still, the number of patients and devices studied thus far is limited and further work is needed to develop manufacturer protocols for establishing conditional CMR safety of pacemakers and ICDs. Carefully designed protocols for patient selection, monitoring, and scanning also need to be developed before imaging of patients with devices can be more routinely performed. The impact of device related artifacts on cardiac image interpretation also needs to be more carefully studied.