MRI-guided endovascular intervention: current methods and future potential

Abstract

Introduction:

Image-guided endovascular interventions, performed using the insertion and navigation of catheters through the vasculature, have been increasing in number over the years, as minimally invasive procedures continue to replace invasive surgical procedures. Such endovascular interventions are almost exclusively performed under x-ray fluoroscopy, which has the best spatial and temporal resolution of all clinical imaging modalities. Magnetic resonance imaging (MRI) offers unique advantages and could be an attractive alternative to conventional x-ray guidance, but also brings with it distinctive challenges.

Areas covered:

In this review, the benefits and limitations of MRI-guided endovascular interventions are addressed, systems and devices for guiding such interventions are summarized, and clinical applications are discussed.

Expert opinion:

MRI-guided endovascular interventions are still relatively new to the interventional radiology field, since significant technical hurdles remain to justify significant costs and demonstrate safety, design, and robustness. Clinical applications of MRI-guided interventions are promising but their full potential may not be realized until proper tools designed to function in the MRI environment are available. Translational research and further preclinical studies are needed before MRI-guided interventions will be practical in a clinical interventional setting.

1.0. Introduction

The main therapy for many vascular diseases, along with drug treatment, is image-guided endovascular intervention. Endovascular interventions are typically performed by inserting a catheter into the femoral artery, which is then navigated under x-ray guidance through the vasculature to the treatment site. The diseased area may be a portion of the vessel wall that has weakened to form an aneurysm susceptible to hemorrhage, or a vessel may become stenotic or occluded by a plaque or clot. Vessels can also become sites of abnormal growth in the case of vascular malformation. The endovascular intervention may involve catheter-based delivery of an embolic material or therapy drug, or deployment of a medical device, such as a detachable aneurysm coil or stent to allow blood flow through a stenotic vessel. Furthermore, another important treatment application lies in oncology, in which chemotherapy drugs or radiation-emitting particles can be delivered directly to the vascular bed feeding a tumor. Minimally invasive procedures, such as endovascular intervention, continue to replace invasive surgical procedures because of their effectiveness with reduced mortality, morbidity, and strain on hospital resources [1].

Most endovascular interventions today are guided by x-ray fluoroscopy because of its temporal and spatial resolution [2,3], utilizing flexible catheters with optimized mechanical properties. The concept of utilizing magnetic resonance imaging (MRI) in interventional procedures began in the late 1980s, where the focus was on MR-guided biopsies [4–6]. In parallel with this innovative work, plastics for catheters and stents were beginning to be identified to expand the arsenal of MR-compatible and visible materials [7,8], and interest in endovascular interventions began to increase [9,10]. Since then, the field of interventional radiology has adopted various new technologies in MR imaging and medical devices to improve treatment efficacy of image-guided endovascular interventions. MRI offers unique advantages for image guidance, but also brings with it different challenges. MRI provides superior soft-tissue contrast, has no ionizing radiation, and allows for measurement of quantitative biomarkers, making it an attractive alternative to x-ray [11,12]. However, several challenges have limited the widespread adoption of interventional MRI, including lack of fast, real-time imaging, scanner utilization, device visualization, MRI safety, and interventionalist ergonomics.

In this paper, we discuss the basic features and current status of MRI-guided endovascular interventions in terms of the MR imaging system, interventional equipment and devices, procedure characteristics, and personnel involved. We will first give an overview of the current state of MRI-guided endovascular interventions, and then speculate on future potential about each category, providing the rationale and evidence for each prediction.

2.0. MRI system

2.1. Magnet and suite configurations

Early low-field MRI systems became available in various formats to facilitate patient access during intervention. The first dedicated interventional MRI, the Toshiba ACCESS, was a 0.064T open system with 360° horizonal “temple” access [13]. Other early designs included C-arm shaped 0.2T Siemens Magnetom OPEN and 0.23T Picker Outlook [14] and additional open systems like the 0.3T Hitachi Airis. These initial systems provided significant horizontal access, but full patient access was suboptimal [15]. The GE Signa 0.5T “double donut”, was then conceived to allow flexible vertical and horizontal patient access through a vertical gap between two superconducting rings [16].

In other approaches, conventional systems have been adapted for intraoperative scanning. Careful attention must be paid to the integration of surgical and MR environments, requiring a multidisciplinary approach among groups including relevant medical specialties, hospital engineers and physicists, and facilities planning, to name a few[17]. The University of Minnesota Intraoperative MRI suite with Philips employed a mobile operating table, permitting rotation outside the 5-gauss line for neurosurgical operation with MRI-compatible tools and into the magnet for imaging [18]. The 1.5T Philips Gyroscan ACS-NT also featured 100-cm tapers on the front and rear, enhancing patient access with two potential surgical areas. Additionally, the Siemens Brain Suite, in collaboration with Brain LAB, implemented a similar rotating operating table in conjunction with a conventional 1.5T scanner [19,20].

The aforementioned MRI scanner configurations have seen suboptimal utilization because of their location in a sterile environment and the lack of MRI scanning being performed during operation. As a result, a clear need has surfaced to perform diagnostic MRI scans independently of lengthy operations. Concepts of movable scanners in dual-, or multi-room, designs have been presented to maximize operational usability of scanners for both diagnostic and surgical purposes in this way. First reports using a single-room movable 1.5T MRI system [21,22], and later dual-room solutions with a movable magnet have been proposed [23]. In these formats, the scanner can move into an operating room, either or via ceiling rails (IMRIS, Winnipeg, Canada), or as a mobile unit [24,25]. Various multi-room arrangements exist with multimodality imaging capabilities at the present. At Yale University, there is a hybrid IMRIS system with a ceiling-rail mounted 3.0T MRI scanner and biplane angiography system. The interventional cardiac MRI suite at the NIH has a 1.5T Siemens Sonata scanner, including a screen, shielded projector, and headphones. A Siemens Axiom Artis FC single-plane angiography system is in an adjacent room, where there is anesthesia equipment, lights, large displays, and an injector [26]. The two imaging suites are separated by RF-shielded doors to permit independent use of each system or simultaneous use with a motorized transfer table. At Brigham and Women’s Hospital (BWH) and Harvard Medical School, the Advanced Multimodality Image-Guided Operating (AMIGO) suite has three connecting rooms that include an interventional PET/CT scanner, an x-ray fluoroscopy table and C-arm, and a 3.0T MRI scanner (Figure 1A) [27]. At University of California, San Francisco’s affiliated Zuckerberg San Francisco General Hospital, there is a fully integrated two-room x-ray angiography and MRI (AMR) suite (Figure 1B). The AMR suite is equipped with a full-featured Siemens Artis Q biplane angiographic system and a full diagnostic Siemens Magnetom Skyra 3.0T MRI scanner [28]. These multimodality interventional imaging suites function as test beds for advanced interventions, enabling independent use of the best features of each type of imaging during different phases of an operation.

Figure 1.

Current interventional MRI suites often combine with other imaging modalities. a) The Brigham and Women’s Hospital’s AMIGO Suite combines a PET/CT, x-ray fluoroscopy table, and a 3.0T (image from https://ncigt.org/amigo), b) Rendering of Zuckerberg San Francisco General Hospital’s AMR Suite Magnetom Skyra 3.0T MRI and Artis Q biplane angiographic system (image courtesy of Siemens Healthcare GmbH).

2.2. Ancillary accessories

A standard x-ray angiography suite typically includes a number of equipment units, multiple operators, imaging control, image display, easy communication in a quiet room with patient and staff, nurses, anesthesiologists, and devices. Standard equipment used in angiography suites often are not compatible for use in the MRI suite, such as anesthesia pumps and ventilators, and oxygen tanks. While there are commercially available MRI-conditional versions, user interfaces may vary, underscoring the need for proper training and in-servicing of MRI-conditional equipment prior to use. The staff needs to be comfortable with and trained on the different interfaces and functionality of the MRI-safe and compatible equipment. MRI-conditional equipment must be clearly marked to ensure that incompatible, potentially dangerous equipment is not introduced to the MRI environment.

For real-time navigation, in-room LCD displays have been developed by Cambridge Research Systems and NordicNeuroLabs to be compatible with 1.5, 3.0, and 7.0T environments outside the Gaussian lines. Communication within the MRI suite between the patients, in-room staff, and control room staff must also be taken into consideration. Standard MRI sequences are loud and require personnel to wear headphones during scans. One solution is noise-cancelling wireless headsets with multiple transceivers around the suite using diffuse infrared technology, with multiple channels to allow movement and communication (IMROC IR^™ Wireless Communication, OptoAcoustics, Tel Aviv, Israel). These headphones would be of particular use when the patient is under conscious sedation, where communication with staff is critical. MRI vendors also now offer quiet imaging modes, which may be applicable to the interventional setting provided the imaging requirements are not compromised in a substantive way.

2.3. Real-time imaging and guidance

Diagnostic MRI has off-line reconstruction and fixed scan protocols. For real-time imaging, reconstruction must be done within a few hundred milliseconds, inline reconstruction with low latency. The introduction of the sliding window technique was essential in allowing MRI to reach real-time frame rates, allowing for the beginning of “MR fluoroscopy” [29]. The sliding window approach initially collects an image with full k-space coverage. The acquisition continues and previously acquired k-space lines are replaced with the most current data. Thus, images can be acquired more rapidly than the time it takes to acquire the full k-space data. The image update rate is then defined by the window of new data that is acquired before an image update is obtained. With this approach an image display rate of 12.5 images/second was achieved with a single coil. This first work resulted in image acquisition time of 250–1000 ms, imaging rate as high as 30 Hz, and a reconstruction time of 67ms [29]. This could be sped up by upwards of 8 times if the approach is parallelized, in combination with larger coil arrays [30,31].

One major challenge in the interventional MRI suite is that personnel and image displays need to be inside the scan room, compared to diagnostic imaging, where there are fixed protocols, image review is done off-line, and no personnel are in the scan room. During an endovascular intervention, the interventionalist needs interactive scan control patient-side, real-time acquisition and reconstruction, possibly hemodynamic recording for cardiac applications, and ability to scan devices. MRI safety training is imperative. Part of the interactive scan control requires being able to change parameters. For the real-time display interfaces, the MRI system vendor’s have their own platforms, and several third-party platforms have been developed, including HeartVista’s research platform, RTHawk [32], OpenIGTLink [33], and VURTIGO [34]. Recently, one research group developed a real-time acquisition and reconstruction system using mixed-reality 3D display for cardiac MRI-guided procedures [35], demonstrating that mixed-reality could potentially be used for real-time intervention in the future.

2.4. Robotic assistance systems

Within typical high-field, closed-bore MRI scanners, patient access is limited, presenting uncomfortable and unergonomic conditions. Robotic-assisted systems have been developed to address the issue of patient access without sacrificing physician ergonomics or image quality. In fact, robotic systems have been realized to function in the MRI environment for multiple surgical applications, yet dedicated systems for minimally invasive procedures are less common and have yet to find market success. This could be partially due to the material limitations presented by MRI, as metallic, ferromagnetic materials must be avoided to maintain sufficient image quality and MRI-conditional status. The field of MRI-guided robotic interventions continues to see innovation with many benefits including improved precision and stability, less interventional trauma, and shortened patient recovery times [36].

Robotic systems designed to augment MRI-guided needle placement is an active area of research, but none have yet to become commercially available. For example, the GantryMate (Interventional Systems, Kitzbühel, Austria) is an MR-compatible system constructed from plastic and fiberglass materials and Bowden cables are used to translate mechanical manipulation [37]. More sophisticated roboticized systems with various actuators are used to remotely manipulate working arms from the control room. The ArthroBot is a remote-controlled, body-mounted robot operating with 4 degrees of freedom (DoF) for MRI-guided needle placement during shoulder arthrography [38]. Similarly, the PainBot features the same 4 DoF system, with an added 2 DoF needle driver system to provide rotation and translation from a driver box for perineural injections. As presented by Cleary, et al., an MRI-safe robot was developed with 3 DoF for pediatric long bone biopsy, using pneumatic actuation and MRI-safe construction materials [39]. Furthermore, the robot controller is in the control room, and data encoding is entirely light or optical, presenting virtually no interference.

While there are commercially available patient-and-control room-side operated systems for endovascular interventions under x-ray guidance, such as the CorPath GRX platform (Corindus, a Siemens Healthineers Company, Watham, MA, USA), equivalent systems for MRI are not yet available. Kundrat, et al., have designed a novel MRI-safe robot for master-slave remote manipulation of guidewires and catheters [40], which may pave the way for future MRI-guided interventions. The robot operates with valve-based pneumatic actuation in 6 DoF, and the physician may easily switch the driver mode between the catheter and guidewire. The slave robot and pneumatic tubing are single-use disposable products, eliminating the need for sterilization between procedures.

3.0. Devices for MRI-guided procedures

3.1. Safety considerations

Fundamental design of interventional devices is important to understand for safe translation into interventional MRI procedures. Typical guidewires and catheters used for guiding interventions under x-ray fluoroscopy feature metallic components in their construction. In the case of guidewires, there is usually the core wire, tip, body, and coating material [41], with the core wire contributing most of the mechanical characteristics. On the other hand, catheters primarily utilize a reinforcement layer that is coiled, braided, or a combination of the two, sealed between a lubricious inner liner and a thermoplastic outer jacket. Like the core wire of a guidewire, the reinforcement layer of a catheter confers most of the device’s mechanical characteristics.

In both guidewire cores and catheter reinforcement, metallic alloys are employed on a device basis for optimized mechanics such as controlled flexibility, pushability, torque response, and kink resistance. They can be stainless steel and nitinol (Nickel-Titanium) alloys in the case of guidewire cores, and catheters braids may also use 304 V, 304 L, and 316 L stainless steels or tungsten [41]. However, due to multiple risk factors in the unique working environment of MR suites, metal components in interventional devices are greatly limited. Stainless steel may not be compatible with MRI scans, as it has variable metal composition and can be ferromagnetic, introducing risk of force and torque from the static magnetic B₀ field. Nitinol, a widely used shape memory alloy, is not ferromagnetic, thus not susceptible to magnetic-induced torque or displacement. These metals function as long conductive structures in braids and wires, interacting with the RF pulses and coupling with the electric field of the transmit RF coil, resulting in elevated power deposition and local specific absorption rate (SAR) at the device tip [42,43]. A major safety hazard arises from this coupling with the transmit RF field, as tissue immediately adjacent to the device tip can undergo unsafe heating during a scan. In one classic paper, a Terumo standard angiography guidewire, consisting of nitinol, was partially immersed in a saline bath to simulate an endovascular intervention. During a gradient echo (GRE) sequence with an average SAR of 3.9W/kg at 1.5T, the distal tip increased from room temperature at 26°C to 74°C after only 30 seconds of scanning [44], demonstrating the danger from not only the heating, but also the very short amount of time required to reach that temperature.

It has been found that conductive components longer than a quarter wavelength at resonant frequency are subject to standing wave formation and RF-heating. Yeung et al. determined that significant SAR gain occurred when the length of a conductive component was longer than a quarter wavelength at resonant frequency, which was approximately 12 cm in humans at 1.5T[45]. By maintaining component lengths shorter than a quarter wavelength, standing wave formation can be avoided and RF-induced heating is mitigated. Multiple groups have identified additional manufacturing steps to achieve MR-compatibility when using metal components. For example, Basar, et al., used a segmented nitinol guidewire for use at 1.5T with connectors of similar mechanical performance in the gaps to eliminate intermittent areas of mechanical weakness [46]. Adapting a similar approach with laser segmentation, Yildirim et al. demonstrated catheters for MRI-guided cardiac interventions through segmentation of a nitinol braid to mitigate device heating without significantly compromising mechanical properties at 1.5T [47]. These manufacturing techniques appear to preserve mechanical performance without the cost of unsafe device heating, but comprehensive mechanical characterization, as required for regulatory approval, has yet to be performed. More importantly, segmentation parameters much be refined and adjusted on a scanner basis, as non-resonant segment lengths would shorten at higher field strengths.

While polymers may not possess ideal elasticity, there may still exist alternatives to metallic reinforcement that have sufficient mechanical performance in specific use cases. Like metallic materials, shape memory polymers utilize high glass transition temperatures to permit heat setting of various desired tip shapes. In another approach, polymers can be strategically selected such that their recovery glass transition temperature is at body temperature (37°C) [48]. These properties allow for devices to be introduced with high stiffness while softening during navigation within the body. Promising polymer candidates have been identified such as para-aramids, liquid crystal polymer, and poly(ether) ether ketone [49,50]. As with the introduction of new manufacturing methods, full mechanical characterization is necessary to support regulatory approval. For interventional devices, common failure modes resulting from insufficient reinforcement include severe bending and kinking, fracturing, separation, and insufficient torque response. In order to obtain regulatory approval, catheters and guidewires must be thoroughly characterized per ISO 10555–1:2013 and ISO 11070:2014, respectively [51,52]

Advances in device engineering have made great strides in driving interventional MRI procedures closer to clinical practice, with preliminary studies completed in swine [53–55]. Before this, device safety bench tests including force and torque from the main static B₀ field and RF-induced heating from the B₁ field should be performed. At a minimum, devices must meet the ASTM standards for MRI-conditional status, which may require phantom testing and simulations [56–58]. Second, for devices to be used clinically, they must also comply with ASTM and ISO standards for sterility and durability. After further development, the next step will be to perform in-human clinical trials and demonstrate non-inferiority when compared to conventional x-ray fluoroscopy devices.

Safety protocols have been established for interventional MRI procedures [59], but they must be revisited with the introduction of new technology. Training is essential to incorporate proper safety checks into the workflow of the surrounding area and the MRI suite. Personnel entering the MRI suite should be checked for metallic materials. Prior to entering the MRI suite, the patient must be checked for potentially dangerous metallic materials, such as EKG electrodes or pacemakers. Equipment entering the MRI suite, such as anesthesia pumps, ventilators, and gas tanks, must be at least MRI-conditional and clearly marked as such in accordance with ASTM standards [60,61]. If equipment is not safe to enter the MRI suite, tubing and wires may be routed through RF-shielded wall conduits to the patient, introducing additional layers of complexity to interventional MRI procedures.

3.2. Guidewires

Two companies have designed guidewires that use a fiber-reinforced material instead of metal braiding for enhanced MRI safety and visualization. The first-in-man intervention was performed using a nitinol-tipped guidewire with iron-doped fiberglass passive markers for patients with valvar pulmonary stenosis [62]. MaRVis Interventional GmbH has obtained CE Mark for their guidewire, MaRVis Amber Wire, which comes in several different diameters, types, tips, and lengths [63]. It is constructed of glass or aramid fibers impregnated by epoxy resin, with a core rod containing iron oxide nanoparticles to induce susceptibility artifacts down the length of the shaft [53,64]. Nano4imaging has a guidewire with MRI conditional status, the MRWire EmeryGlide^™, with CE mark and FDA 510K approval for use at 1.5T and 3.0T. It also contains discrete markers, and the length of the wire is radiopaque [65]. While one approach might be to modify the reinforcement materials of a guidewire, there are a number of previously proposed solutions to mitigate RF-induced heating, including various RF transmit solutions [66–68], device modification [69–72], and sequence modification/power limits [73,74]. One relevant example of sequence modification and power limits was recently demonstrated, where the investigators were able to limit the heating of a specific fully insulated metallic nitinol guidewire (150 cm, 0.035” Terumo Glidewire) to an increase of just 0.07°C, using a low-SAR imaging sequence at 1.5T [75]. The imaging protocol used a reduced flip angle (10°) GRE acquisition, with a longer repetition time (TR) spiral readout (10ms) for cardiac MR (CMR) fluoroscopy. The authors emphasized that this needs to be verified for each scanner, as certain parameters may change between different vendors. At lower field strengths (i.e. 0.55T), RF heating of guidewires becomes less of a concern, at the tradeoff of reduced SNR and lower polarization [76]. Lastly, parallel transmit (PTx) coil arrays have recently demonstrated safe visualization of unmodified guidewires [77]. In this study, current sensors were placed over a standard guidewire to determine RF-induced current modes during interventional CMR. The PTx system uses maximum and null current modes to safely image with a guidewire in situ; however, this approach has noticeable signal drop-off at the distal tip.

3.3. Catheters and MRI-visible markers

MRI visualization markers for catheters or guidewires are typically categorized as passive, active, or resonant (also referred to as semi-active) [78]. Passive approaches use differences in magnetic susceptibility between the device and surroundings, such as gadolinium or iron oxide particles, to cause local B₀ field inhomogeneities, giving rise to signal voids [79,80,9,81]. Their visibility may suffer from partial volume artifacts and diminished contrast in the image. One of the first implementations was real-time MR fluoroscopy of an MRI-guided iliac artery stent placement using dysprosium (Dy) doped catheters (Figure 2) [82]. An early preclinical application of a passively-tracked catheter used a gadolinium (Gd) filled balloon for iliac artery stenosis in animal models [83]. Another example was a CO₂-filled balloon for cardiac catheterization in patients to measure pressure [84]. These devices were then able to demonstrate feasibility of pulmonary vascular resistance using pressure measurements combined with blood flow measurements [85], demonstrating improved accuracy than the gold standard. This was further extended in a larger patient study, allowing accurate stratification for intervention in patients with congenital heart disease [86]. MRI-guided diagnostic cardiac catheterization is now regularly performed in patients at the NIH using passive air or gadolinium-filled catheters [87].

Figure 2.

MRI-guided iliac artery stent placement using Dy-doped catheters. a) Visualization of Dy markers (arrows) along the nonmetallic wire. b) The catheter-mounted ZA stent (arrows) positioned in the aorta c) The stent withdrawn to the position of deployment, and d,e) Stent deployment in real-time. f) Fully deployed stent (arrows) is shown [49].

Dy = Dysprosium

Active markers, such as micro receive coils, may supply a direct current (DC) to a coil on the catheter [88–90], or in another approach, may use miniature transmit-receive coils [91]. Active tracking has been achieved with MRI needles featuring an elongated loop coil and multiple tight windings with passive guidewires [92], and more recently, a single-loop coil with multiple solenoid coils printed in conductive ink down the shaft of a nitinol-based needle [93]. First in-human feasibility studies of actively tracked catheters were demonstrated with interactive software (iSuite, Philips Healthcare, Best, Netherlands) to display an electrogram measuring the isochrone for an electrophysiology procedure to ablate an atrial arrhythmia [94,95]. This work has been built upon to develop a portfolio of MRI-conditional steerable diagnostic and ablation catheters (Imricor, Minneapolis, MN, USA), which recently received their CE marks. Active tracking is challenged by the requirement to supply external current via conductive wires down the length of the catheter, which are susceptible to local RF-induced heating at the device tip [44]. Moreover, transmission lines occupy precious space in the small working channel, making manufacturing of interventional devices with decreasing diameters a considerable challenge. Studies have sought to mitigate heating by utilizing high-resistivity coaxial cable [96] and modifications such as coaxial chokes to reduce undesirable currents. Recent work has used acousto-optic markers, using optical fibers for signal transmission at 0.55T [97]. There is still room for improvement to adapt these techniques to higher field strength scanners and smaller devices.

Wireless resonant circuit markers, or inductively coupled RF coils, can be implemented as a circuit with a capacitor (C) connected to a coil, which has inductance (L) and intrinsic resistance (R). This RLC circuit can be built on the distal tip of the catheter to provide local signal amplification of the B₁₊ field [98,99]. The circuit inductively couples with the RF transmit coil, inducing current, which is then amplified through the resonant circuit. This causes the local RF field to increase, resulting in local signal amplification and a localized bright signal at low input RF levels. Such wireless resonant circuits [100], also known as inductively coupled RF coils [101], have previously been built on top of large diameter guide catheters using a variety of fabrication methods and geometries. These wireless circuits eliminate the need for wires running down the length of the catheter for visualization, ultimately reducing heating risks that plague active tracking methods. Wireless resonant circuits have been built more recently on catheters for interventional MRI, using inductor coil geometries including rectangular surface loop coil [102], two opposing solenoids oriented at 45° relative to the major axis of the catheter [103], and flat spiral coils [104]. Recent work proposed a hand-wound circuit with an inductor geometry of a double helix, with each arm tilted 45° with respect to the z-axis [105,106], which was then 3D-printed, with feasibility shown in vitro and in vivo [107] (Figure 3). As with active tracking, resonant approaches may suffer from bulky components, leading to large device size and suboptimal device rigidity. As a result, a number of other microfabrication methods for building markers on catheters have been explored beyond manual wire winding, including flexible printed circuit boards, hot embossing, thin-film techniques, and aerosol deposition processes [108]. One group designed and manufactured rectangular spiral planar coils stacked in multiple layers, with a size of 6.7×1.5×0.3 mm³, built on flexible printed circuit boards, and incorporated it onto needles and catheters [109]. The aforementioned techniques remain in investigative phases of feasibility and material property exploration with limited in vivo data.

Figure 3.

Wireless resonant marker tracking in vivo at 3.0T. Markers placed in the right (dashed arrow) and left (solid arrow) carotid arteries of a single swine are shown in a coronal scan plane under a) MRI and b) x-ray, which are more easily seen under c) magnification (left artery with contrast, and the right artery without contrast). d) A sagittal B1+ map quantifies signal amplification. Low background signal and high marker signal amplification was demonstrated at a flip angle of e) 5°. Surrounding tissue signal becomes higher and over-flipping near the marker are shown as the flip angle increases to f) 10° and g) 45° [74].

3.4. Steerable/direct current catheters

Visualization is one important aspect for navigation, and other forms of navigation assistance have manifested in steerable catheters. These devices offer the ability to torque the distal tip through some of the most tortuous anatomy [110]. Manual actuation was demonstrated by Clogenson et al., where the distal tip of a multi-selective catheter was actuated via pull-wires [111]. Gosselin et al. exploited the gradient forces during MRI scanning to steer a catheter tipped with ferromagnetic spheres [112]. These spheres ultimately could cause unintended displacement, arising from dipole-dipole interactions. Other potential limitations that could make navigation difficult include image artifacts induced by the spheres and the physical weight of the tip causing traumatic shear stress on vessel walls [113].

Steering can also be achieved via remote controlled catheters actuated by applying direct current to coils at the distal tip. An induced magnetic field at the catheter tip then interacts with the static field of an MRI scanner, allowing on-demand application of field-aligning torque. Lillaney et al. demonstrated renal embolization in a swine model at 1.5T [114]. As with active tracking methods, long wires supply current to the steering coils, which were susceptible to heating and melting of insulative jacket materials. Later work determined safe upper current limits to avoid such heating [115]. Liu et al. modeled a magnetically-actuated steerable catheter in three-dimensional space [116], and improved upon it using open-loop control [117], building on prior work from Guidino et al. with an array of active steering coils [118]. Early iterations of steering coils may lack the low profiles needed for endovascular intervention, resulting in rigid and cumbersome devices that are unnavigable (i.e. do not “track”) in finer vasculature, making advanced manufacturing methods (i.e. laser lithography) important for low-profile technology.

3.5. Tracking sequences

Active tracking is achieved with tracking sequences that use intermittent non-selective excitations combined with spatially encoding gradients [119]. The first MRI tracking sequence of an active marker, a micro-coil, used four projections to acquire the coil position [88]. Next, continuous radial MRI was introduced for real-time simultaneous acquisition of anatomical MRI and tracking of micro-coils [120]. With additional distributed memory capability, any arbitrary k-space trajectory was then possible to use for real-time MRI [121]. Then, bSSFP sequences were implemented real-time [122]. Later, Hadamard encoding and phase dithering were added to improve the active catheter tracking sequence [91]. This was further implemented for use in tracking wireless resonant circuits [100]. Another real-time active tracking method added a tracking block to a bSSFP and spoiled gradient recalled echo (GRE) sequence, and repeated the block four times with different readout gradient polarities to correct for off-resonance [123].

Several groups have developed real-time or near real-time MRI sequences for use in conjunction with a passive tracking strategy. Bakker et al. proposed contrast-enhanced MR fluoroscopy as a tracking method in conjunction with passive catheters and guidewires [9]. Martin et al. developed a digital subtraction MR angiography roadmap to track steerable catheters with passive contrast [124]. Another method tracked a catheter with a passive ferromagnetic tip using dual-echo projections [125], by using off-resonant excitation and rephasing gradient to selectively excite and conserve the signal around a sphere. To remove the background signal and preserve the signal in the selected region, two projections are acquired in one echo train with orthogonal direction of the rephasing gradients. There have also been developments of new visualization methods in combination with tracking, such as white marker imaging, which make the passive signal dropout of a passive marker turn bright (Figure 4) [126]. Patil et al. developed automatic slice positioning for passive real-time tracking of devices [81], by using their susceptibility shape and dephasing gradients interleaved in sequence. Concurrent contrasted-enhanced visualization of patient anatomy and gadolinium-filled balloon catheters has been achieved by Forte et al. through partial saturation (pSAT) sequences comprised of a single shot acquisition with bSSFP readout preceded by a partial saturation pre-pulse [127]. Overall, passive tracking often requires specific pulse sequence modifications tracking to avoid partial volume effects by imaging across thicker slices, but contrast to noise may be sacrificed in the process.

Figure 4.

Passive marker tracking sequences. Visualization of three paramagnetic markers with a) a conventional gradient echo sequence (slice 30 mm, TE/TR = 4.6/60 ms, duration 22 sec) and b) dephased positive contrast gradient echo imaging (white marker sequence with 1.9 cycles of phase across the slice) with similar acquisition parameters. In vivo application of white marker tracking with significant obscuring of the makers for conventional c) unsubtracted and d) subtracted tracking. White marker tracking permits easy detection of the markers for both e) unsubtracted and f) subtracted positive contrast tracking [93].

4.0. Clinical applications

4.1. Cardiovascular applications

Promising cardiovascular applications include not only identification and treatment of coronary atherosclerosis but also evaluation and treatment of congenital heart disease and electrophysiology [119,128–130]. Razavi et al. performed MRI-guided cardiac catheterization in sixteen patients with congenital heart disease for purposes of diagnosis and evaluation, including two radiofrequency ablations [84]. The elimination of ionizing radiation was particularly meaningful in these cases, since they were performed on children and teenagers [84]. Performing these procedures under MRI has the added benefit of permitting any slice orientation. Recently, Campbell-Washburn et al. performed right heart catheterization using a metallic guidewire, and using a low SAR sequence, measured negligible heating (<0.07°C) using a standard angled-tip nitinol Terumo Glidewire [75]. Ratnayaka et al. performed 50 cardiac MR fluoroscopy guided right heart catheterizations in 39 pediatric patients, using passive catheters [131].

In the pilot studies outlined above, the standard fluoroscopic procedure was largely copied directly to MRI to demonstrate feasibility; however, they have yet to show a clear advantage of real-time MRI guidance beyond the avoidance of exposure to ionizing radiation. Further modifications will be needed before any endovascular MRI-guided procedure can be adopted for more widespread clinical use.

In the future, more difficult cardiac procedures, such as treatment of ventricular tachycardia could benefit from MRI-guided procedures. Another compelling MRI application is measuring force, as well as ablation area and temperature, and correlating with the late gadolinium enhanced images (trans-septal puncture) [129,132,133]. Real-time MR thermometry has also been used to quantitate thermal dose during real-time ablation, which was recently demonstrated and implemented in Gadgetron: An open-source framework for medical image reconstruction [134]. Moreover, Gadgetron has been shown as a platform for high-resolution spiral imaging to image guidewires navigating into the heart [135,136].

4.2. Peripheral arterial applications

MRI guidance to perform endovascular procedures offers the possibility to quantitatively measure success in real-time while performing the procedure without having to move the patient. Procedures that could benefit from intra-procedural imaging of a physiologic biomarker and superior soft tissue contrast are of particular interest. For example, pelvic, femoral, and popliteal artery stenosis have been stented using real-time endovascular MRI [137,138]. Manke et al. demonstrated feasibility of performing angioplasty and placing stents in thirteen patients with iliac stenosis [137]. Overall, while the procedure was feasible, it was limited by lack of real-time monitoring, stent artifacts and long procedure times [137]. Similarly, Paetzel et al. demonstrated real-time MR-guided balloon angioplasty of femoral and popliteal stenoses [139]. Their research benefitted by lack of stent use (limiting image-based artifact), injection of gadolinium into the angioplasty balloon for visualization, and pre and post intra-arterial MR angiography [138]. This research shows promise, particularly with current advancements in MR imaging techniques, highlighted by the use of MRI to evaluate intra-plaque hemorrhage in high-risk atherosclerotic plaque and quantitate flow [140]. Non-invasive MRI assessment of arterial plaque morphology could help guide intervention in the future [128,140].

4.3. Oncology applications

4.3.1. Central nervous system oncology

MRI has been adopted over other modalities for diagnosis and management of multiple oncologic diseases because of superior soft tissue contrast and the ability to obtain multiparametric tissue characterization. Recent research has found value in MRI during osmotic blood-brain barrier opening (OBBBO) to increase treatment efficacy of intra-arterial therapies for primary and metastatic disease in the brain. The current gold standard for such procedures is x-ray digital subtraction angiography (DSA), though it has yet to gain popularity despite being introduced over 4 decades ago[141]. Primary reasons behind the lack of traction could be outcome variability and a lack of real-time validation of blood-brain barrier permeability. Initial studies have shown that dynamic contrast susceptibility (DSC) MRI defines catheter perfusion territory and contrast-enhanced images delineated OBBBO territory [141]. A first-in-human study by Zawadzki et al. demonstrated safety and feasibility of real-time MRI guidance for intra-arterial therapy delivery following OBBBO [142,143]. Authors emphasized that superselective IA delivery of therapy, in conjunction with real-time MRI guidance and validation, decreased volume of the enhancing mass, rapidly improved neurological status, and potentially improved survival in a rapidly declining patient with glioblastoma [142].

4.3.2. Liver oncology

Respiratory and cardiac motion affect abdominal organs to a greater extent than the central nervous system, requiring high temporal resolution. MRI can be used to ascertain not only spin thermal relaxation such as T1 and T2 relaxation times, and molecular electron cloud polarization, such as magnetic susceptibility, but also contrast enhancement kinetics and water motion, such as perfusion and diffusion properties in one setting. MR monitoring of intra-arterial contrast infusions has demonstrated the ability to clearly demarcate tissue fed by the selected artery, providing insight into the impact of embolics administered via that vessel [144]. These advantages have prompted some groups to investigate use of MRI during endovascular oncologic treatments such as chemoembolization or radioembolization [145,146]; however, clinical applications have thus far mostly used coregistration of preprocedural MRI scans to intra-procedural cone beam CT. Nevertheless, coregistration of preoperative MRI to intraprocedural imaging is subject to errors such as misregistration, changes in the patient or disease process between the two scans, and variation from respiratory or cardiac motion. The development of MRI-conditional equipment could greatly aid in performing such interventions with real-time MRI guidance. One group has used intraprocedural transcatheter MRI perfusion imaging to assess chemoembolization endpoints in an angiography-MRI suite; though, the key procedural steps such as target vessel catheterization were performed using x-ray fluoroscopy [147]. As MRI biomarkers of treatment success become more established as surrogate markers of patient outcome, the advantages of MRI guidance for oncologic treatments will only increase.

4.4. Preclinical applications

Many endovascular interventional radiology applications have been tested in vivo in animal models, given the lack of clinically approved devices. Notable preclinical applications that could benefit from using MRI to quantitatively measure success are stroke embolectomy and tumor embolization. Recently, one group demonstrated a proof-of-concept performing a carotid embolectomy under MRI guidance in a swine stroke model [148]. This approach could be particularly valuable in acute ischemic stroke where intervention could be streamlined into one interventional suite, reducing treatment delays and allowing for intra-procedural evaluation of brain parenchyma viability, such that reperfusion therapy could be directed to living tissue and not at infarcted tissue [149]. Lillaney et al. demonstrated the ability to embolize renal arteries in a swine embolization model, additionally evaluating perfusion and flow pre- and post-procedure as imaging biomarkers [114]. Similar studies have been executed demonstrating renal artery embolization and hepatic artery drug infusion [150,151]. This approach could be particularly valuable in tumor embolization where new MR imaging techniques are being applied to evaluate tumor perfusion reduction during chemoembolization and have been shown to predict transplant free survival [147]. Other interventional radiology applications that have been tested in animal models include endovascular stenting of aortic aneurysms [152,153], stenting of descending thoracic dissections [154], angioplasty and stenting of arteries including carotid, renal and iliac arteries [82,155–157], inferior vena cava filter placement [158], and creation of porto-systemic shunts [159,160]. Preclinical interventional cardiology applications include coronary artery septum placement [161,162], pulmonary artery stenting [163], balloon angioplasty of aortic co-arctation [164], aortic valve placement [165] and septum occlusion [166,167]. Oncology applications have established small and large animal models for stem cell delivery to the CNS [168] as well as optimization of OBBBO with other intra-arterial therapies [141,169–171]. Finally, researchers have found an intersection between molecular MRI and endovascular interventions with chemical exchange saturation transfer (CEST) MR contrast agents. Preclinical models to better assess OBBBO perfusion territory [172] and function as “label-free” theranostics following brain ischemia [173] have been identified in this field of research in addition to an expanding portfolio of MR contrast agents on the horizon[174].

5.0. Cost

Relatively little attention has been paid to this critical issue. Hall et al. explored the costs and benefits of MRI-guided brain tumor resections [175], concluding the potential costs of interventional MRI could be offset by improved health outcomes. Their work demonstrated that adults who had a brain tumor resection in an intraoperative MRI had a 54.9% shorter hospital length of stay compared to those who had resection performed in a conventional OR, showing a higher turnover rate and financial advantage to using MRI. Currently, the cost of a diagnostic MRI exam is typically 2–3 times more than that of a CT or ultrasound, and the increased cost is an area of significant scrutiny. Nonetheless, there have been efforts in recent years for a “Limited MRI” Current Procedural Terminology (CPT) code. This is an MRI scan with only a few essential, focused sequences, for a clinical indication that should have the best visualization on MRI. The reduced overall time in the MRI scan room (~15 minutes including set-up), in combination with the ability to avoid imaging modalities that are less specific, allow the overall cost to be lowered, such that the cost is on par with that of an ultrasound or CT scan [176]. This same concept could be applied to reduce the cost of interventional procedures; however, MRI-guided therapy must first thoroughly prove its cost effectiveness and value as a minimally invasive surgery over open surgery [177]. In body interventional radiology, biopsy and ablation CPT codes used for CT procedures have been applied to MRI-guided procedures [178]. Authors emphasized the importance of patients consulting with billing offices and their health insurance provider to predetermine out-of-pocket expenses or prior authorization requirements. Ultimately, demonstrating improved clinical outcomes, such as shorter length of hospital and rehabilitation stay and reduced disability in acute ischemic stroke, will drive reimbursement strategy and insurance coverage determination. Access to an interventional MRI unit may help other interventional procedures maximize use and demonstrate clinical value as a return on investment.

6.0. Conclusion

Interventional MRI represents the next era of diagnostic and therapeutic biomedical imaging. Significant limitations in current devices designed for use under x-ray guidance have driven development of an array of next-generation capital equipment, catheters and guidewires, and tracking techniques. Promising technological progress has been made to ensure patient safety and enhance MRI-guided endovascular procedures in real-time. As more MRI-safe technology reaches the market, dissemination of minimally invasive techniques throughout various medical specialties will follow at great benefit to operators and patients alike.

7.0. Expert opinion

MRI can provide crucial physiologic information which standard x-ray guidance cannot, such as diffusion and perfusion metrics, hemodynamics, and soft tissue characterization. In particular, the ability of MRI to precisely and confidently discriminate between living and dead or cancerous and noncancerous tissue could yield important benefits to patients if harnessed properly to better define targets during image-guided interventions. Yet, interventional fields of medicine have not transitioned away from x-ray fluoroscopic guidance.

Hesitance to adopt interventional MRI may stem from multiple root causes. Safety concerns arising from introduction of interventional devices into the MRI environment are relevant and will require careful characterization and consideration. Conversely, other trends in healthcare, such as greater availability of MRI-conditional devices and accessibility to MRI in general, may accelerate adoption and proliferation of MRI-safe environments in hospital settings. Low-field MRI scanners have been deployed to the patient bedside for point-of-care applications, which may also be able to provide improved patient access like prior open-bore configurations [179]. Additionally, new hospitals are increasingly being outfitted with MRI scanners within their interventional and operating suite environments. We expect that these trends will only further cement the centrality of MRI in management of many neurologic, cardiac, and oncologic diseases.

Another concern relates to the limited spatial and temporal resolution of interventional MRI sequences, presenting a challenge to device tracking in 3D space. Myriad techniques have been conceived to improve the imaging ecosystem [180], depending on whether an active, passive, or wireless resonant tracking strategy is pursued. Ongoing development of fast MRI pulse sequences, as well as routine high-fidelity coregistration of preoperative MRI with real-time intraoperative MRI, are expected to continue to improve over time and will further streamline procedural care [181].

Development of interventional devices that are safe and conspicuous in the MRI environment will likely shape the way that interventional MRI is adopted in the future. There are a wide variety of endovascular devices available in a physician’s armamentarium for fluoroscopically-guided interventions; the number of tools available for endovascular intervention under MRI guidance will be significantly less, at least initially. Once a broader selection of MRI-conditional and safe devices becomes available, diversification and larger scale utilization of interventional MRI may follow. While there are multiple types of devices under active investigation specifically indicated for interventional MRI, considerable regulatory hurdles exist prior to entering the market. The most straightforward avenue by which new devices can be approved is through substantial equivalence to commercial devices in a 510K clearance. In the case of novel tracking techniques, this will be a major challenge, as equivalent devices are scarce. The arduous regulatory process in the US, (nicknamed ‘Death Valley’ [182]) encourages more iterative design over paradigm-shifting technology, which may delay clinical translation. Therefore, translational research in conjunction with preclinical data will remain paramount in expanding the number of devices and procedures with utility for MRI-guided endovascular intervention. As combination imaging suites become more prevalent, devices with multimodality visualization capabilities could be beneficial but have yet to be realized.

Technological advancements in single-use devices will act synergistically with those in capital equipment. Advanced RF transmission technologies like PTx coil arrays can enable use of standard guidewires, for example [77]. For these systems to be used in humans, on-board active SAR monitoring is needed. This area of research is active, having demonstrated in vivo success of dynamic RF-induced current control. Future steps are being taken to produce current measuring devices in a sterile format to complement pTx on conventional scanners. Robotic assistance may also enhance MRI-guided intervention, mitigating potential occupational hazards for operators. Future systems could further specialize for specific procedures or specialties, potentially with higher precision and cooperative control where there is real-time haptic feedback in conjunction with hands-on interfaces. Recent research suggests this is possible, with a shift towards fully MRI-safe platforms using non-metallic materials and pneumatic actuation.

Still, the most critical hurdle remains in justifying the costs associated with establishing interventional MRI suites. Bringing up an interventional system requires significant investment; thus thorough clinical advantage must be proven. Given the technical barriers and cost to implementing interventional MRI, future studies must go beyond proof-of-concept to demonstrate improved efficacy over other image guidance modalities such as x-ray fluoroscopy, ultrasound, or x-ray CT. By routinely implementing MRI guidance during cardiac electrophysiology intervention, oncologic intervention, ischemic stroke intervention, among other interventions, improved outcomes could further support these claims. These image-guided interventions, in particular, have evolved to become heavily dependent on preprocedural diagnostic MRI to define targets for treatment. If MRI guidance were integrated into the following procedural workflows, treatment could be streamlined with improved results. Perhaps in the future, interventional MRI scanners could see the most use in multi-room configurations where they can be maximally utilized during interventions or in parallel as diagnostic scanners. These suite designs could still greatly vary on a hospital basis depending on the application, surgeon, procedure being performed, and disease processes involved.

We anticipate a combination of factors leading to increased adoption of interventional MRI over the coming years. These include increased accessibility to MRI within healthcare systems, regulatory approval of single-use disposable MRI-conditional devices, clinical demonstration of MRI’s utility for management of cardiac, neurologic, and oncologic disease, continued development of real-time MRI tracking pulse sequences, and development of robotic tools to aid with the ergonomics of image-guided interventions. Financial and regulatory barriers in the healthcare system are expected to continue to limit rapid adoption of any technology, laying emphasis on the need for robust translational research efforts to ensure safety and efficacy.

Article Highlights.

• Interventional MRI possesses unique benefits over standard x-ray fluoroscopy including superior soft-tissue contrast, no ionizing radiation, and measurement of quantitative biomarkers.

• Due to the significant technical effort and expertise needed, MRI-guided endovascular interventions are only recommended for situations in which MRI provides a significant advantage over other modalities.

• Current commercially available interventional devices lack MRI compatibility and safety profiles. Key characteristics must include ability to visualize the distal tip without sacrificing MRI safety and mechanical performance.

• Clinical adoption of interventional MRI requires future generations of MRI-conditional and MRI-safe devices to facilitate better tracking and treatment in real-time.