Most MRI-guided cardiovascular interventions are conducted in 1.5-T scanners with a short wide bore. Three-tesla and low-field open systems are theoretically also suited, but are limited by higher affinity to susceptibility or lower resolution for real-time cardiac imaging, respectively. The short bore of modern scanners allows reasonable access to the groin or neck of the patient, although catheter manipulation can be difficult in small infants or children (Fig. 38.1).

In the clinical setting, x-ray fluoroscopy is used for almost all endovascular interventions, due to excellent spatial and temporal resolutions in combination with high contrast between vessels and background. Recent developments, such as contrast-enhanced 3D rotational angiography, exploit these features further and open up new possibilities. However, x-ray is still limited in terms of soft tissue visualization and acquisition of quantitative functional parameters. Combined x-ray/MRI units were proposed to take advantage of both imaging modalities. Fahrig et al. (3) reported simultaneous x-ray and MRI by integrating, similar to PET-MRI systems, an x-ray camera into an open 0.5-T scanner. A different strategy is to connect the MRI scanner over a mobile table with C-arms or even fully equipped x-ray angiography laboratories (4,5). The use of such combined units makes sense until MRI catheter-tracking techniques have reached a higher standard in terms
of reliability and safety. Until then an x-ray safety back-up system should be available.

The overlay of MRI-derived anatomic roadmaps to x-ray fluoroscopy is used in an increasing number of applications. Such an approach can be applied in combined x-ray/MRI units or, in a sequential approach by importing previously acquired MR images into a conventional catheterization laboratory. Any of such multi-modality imaging requires methods for image registration. For having a kind of overview, registration can be done by anatomical landmarks. For procedures that require more precise information, more sophisticated registration methods need to be applied. Such methods include among others the use of fiducial markers (Fig. 38.2) (6,7).

MRI-compatible equipment for anesthesia and monitoring vital parameters are offered by different manufacturers. In-room display monitors and operation consoles allow image acquisition by an interventionalist and imaging expert standing alongside. An aspect that must be considered when performing MRI-guided intervention is high noise levels during imaging. However, our experience shows that communication can be well adapted to intermittent noise by training of the staff involved in the procedure.

Safety aspects for the use of an interventional MRI unit require that general contraindications for MRI of patients are respected. In addition, the use of interventional instruments must be chosen with their interaction with the magnetic field in mind (8,9,10 and 11). Metallic materials that yield significant magnetic torque, such as introducer needles or wires, must not be brought into close proximity to the magnetic field. In addition, all elongated electrical conductors, metallic or not, are prone to heating effects if they reach the critical length of 20 cm at 1.5 T (10,12,13). Therefore, such interventional instruments must be excluded from en-dovascular manipulation. Further details on the MRI compatibility of catheters, guidewires, and medical implants are provided in other chapters.

Real-time 3D imaging is a task that remains to be realized. However, MRI already offers an exceptional variety of
two-dimensional (2D) real-time and ultra-fast acquisition modalities for the assessment of anatomy and function. Up-to-date high-gradient systems provide the short TR and TE needed for fast imaging with steady state free precession (SSFP). This acquisition technique in conjunction with spiral or radial filling of the k-space is currently considered one of the most reliable tools for high-quality and robust real-time imaging (14,15).

Nevertheless, during real-time MRI the interventionalist has to find a compromise between temporal and spatial resolutions or signal-to-noise ratio (SNR). In-plane resolutions of 1 mm can be obtained if lower acquisition rates of approximately 5 frames per second are accepted. In contrast, at lower spatial resolutions, frame rates of up to 20 frames per second can be achieved.

It is important to note, that x-ray fluoroscopic frame rates of less than 10 frames per second are often applied in order to minimize the radiation exposure of the patient. In addition, MRI allows interleaved high-resolution imaging if needed. The rationale for MRI would be the transition between different imaging sequences that provide either high-spatial/low-temporal or low-spatial/high-temporal resolution.

Accurate assessment of cardiovascular morphology prior to the intervention can facilitate the procedure and reduce its duration. The acquisition of high-quality whole-heart imaging provides detailed 3D information that can be used as an anatomical roadmap when overlaid to real-time images or x-ray in combined units (7,16). Such multi-modality imaging approaches are known for electrophysiology studies but are also promising for closure of ventricular septal defects or performing myocardial biopsy (17).

High-quality MRI of anatomy in combination with MRI-derived blood flow provides valuable information that permit simulating different interventional or surgical treatment strategies prior to the actual procedures. Thus such approaches enable to determine for the individual patient the type of treatment that offers optimum outcome (Figs. 38.3 and 38.4) (16,18).

Multi-modality imaging that fuses MRI or CT images with x-ray fluoroscopy is routinously applied in electrophysiology studies. In a more experimental level, such methods were described for guiding myocardial biopsy and closure of VSD (17). These approaches will be described in more detail in the chapters below.

To date most medical implants used in cardiovascular medicine comprise at least in part metal alloys that can cause artifacts on MR images. Such devices include endovascular stents or stent-prostheses, septal or duct occluders, embolization coils, and prosthetic heart valves. These implants are generally not a contraindication for MRI studies: They yield only minor magnetic torque and heating and therefore can be considered MRI compatible in terms of patient safety (8,19,20). However, susceptibility artifacts of the metallic device can cause pitfalls or even make assessment of cardiovascular morphology and function impossible (20).

The extent of imaging artifacts depends mainly on the paramagnetic properties of the device and the field strength of the magnet. Less important factors are material thickness, the orientation of metallic struts toward the B₀ field direction, and the sequence parameters used during MRI.

Metals with weak paramagnetic properties, such as nitinol, yield susceptibility values that are relatively close to those of human tissue and therefore cause only slight local imaging artifacts (21,22,23 and 24). By contrast, metals with strong paramagnetic properties, such as stainless steel, produce severe image distortion (22,23 and 24). Furthermore, the extent of susceptibility is directly related to the magnetic field strength. MR scanners with high field strengths cause more susceptibility artifacts than scanners with low field strengths (8,25). Finally, parallel alignment of a metallic strut to the B₀ field direction causes fewer artifacts than orthogonal alignment (25,26). However, the orientation of a metallic implant toward the field direction is obviously difficult to predict in the clinical setting.

The MR artifact behavior of endovascular stents is further complicated by radiofrequency shielding effects. Shielding effects are caused by eddy currents on the surface of the stent
that prevent radiofrequency pulses from fully penetrating the wire mesh. This results in decreased signal intensities within the lumen of the stent. The obscured lumen can be misinterpreted as in-stent stenosis or obstruction (21,26,27 and 28).

The impact of pulse sequences for minimizing susceptibility or radiofrequency shielding was investigated in several studies (23,24,27,29,30). The use of large flip-angle can reduce the amount of signal loss by radiofrequency shielding (23,31). Short TE and read recording direction parallel to B₀ were reported to reduce dephasing and thus susceptibility (23,27). A further, quite effective sequence implement is 180 degrees refocusing pulses, as used in spin-echo imaging. However, artifacts on spin-echo images must be carefully interpreted because remaining radiofrequency shielding and residual susceptibility artifacts cannot be easily differentiated from the dark blood pool, and thus in-stent stenosis could be falsely overlooked.

There are a broad range of catheters that are manufactured for specific cardiovascular applications. Some of these catheters are braided with iron oxides or wire meshes in order to make them radio-opaque or to add flexibility. Such catheters

are generally not suited for use during MRI because they can produce severe image distortion due to susceptibility. On the other hand, catheters made from polyesters only do not generate any signal at all and are therefore invisible during MRI. However, appropriate catheter-tracking methods are essential to perform successful MRI-guided endovascular procedures. Ideally, tracking methods generate high contrast between the catheter and the background anatomy and readily allow automated slice and tip detection. In addition, the method should visualize both the tip of the catheter and its shaft in order to be able to identify possible looping of the catheter. The development of catheter-tracking methods that fulfill all clinical needs is technically very challenging. These difficulties are one of the major barriers that delayed the transition of preclinical interventional MRI into the clinical settings.