Image Guidance of Percutaneous Coronary and Structural Heart Disease Interventions Using a Computed Tomography and Fluoroscopy I

Introduction

Several studies have documented the inherent limitations of angiography and X-ray in the accurate diagnosis of coronary artery^1,2 and structural heart disease. Traditionally, invasive coronary angiography has been performed with radiographic equipment that provides a simplistic, two-dimensional (2-D) representation of a patient’s more complicated three-dimensional (3-D) anatomy. This “flattening” of a 3-D image results in the generation of a final image that may be limited and inaccurate. To minimize these limitations, invasive cardiologists typically acquire 6–10 diagnostic views of the coronary arteries aimed at evaluating the multiple vessel segments. In addition to the limitations of intravenous contrast, ionizing radiation, and procedural time, each angiographic view is subjectively chosen to best display an individual patient’s coronary arteries. As it has been previously demonstrated in the coronaries,^2–4 the combination of 2-D imaging limitations and this trial-and-error technique frequently results in inherent and unrecognized imaging inaccuracies. Given the widespread use of coronary angiography, these imaging limitations have important clinical consequences. During diagnostic angiography, a lesion may be missed or the severity underestimated. The consequences of inaccurate imaging during coronary therapeutic procedures include the selection of an incorrectly sized stent, inaccurate placement, and/or the need for additional stents. Echocardiography is used for diagnosis but also for image guidance of many structural heart disease interventions. Suboptimal 2-D ultrasound views may result in the selection of incorrect sizes of atrial septal defect (ASD) and patent foramen ovale (PFO) occluder devices and/or an inappropriate anatomical evaluation of the surrounding structures leading to a poor post-procedural outcome

In an era of expensive cardiovascular treatments, the need for an accurate, efficient, and safe imaging technique is necessary. Three-dimensional imaging resolves many of the 2-D limitations of traditional angiography. Offline 3-D coronary angiography reconstructions have been available and their use validated. Until recently, computer processing requirements and the complexities of reconstructing the dynamic coronary tree have prevented the real-time application of 3-D imaging during angiography. A real-time 3-D imaging tool is now available and validated. Sadly, in the structural heart disease arena, this technology has not been completely developed and is dependent on larger doses of contrast and radiation exposure in order to reconstruct heart chambers. Computerized tomographic angiography (CTA) has revolutionized the noninvasive coronary and structural heart disease evaluation arena. As CT scanners improve and imaging is more reliable, the need for invasive coronary studies has, and will, decrease in certain populations. The validation of CTA is well under way, with promising results.

Our group is the first to extract a centerline-based coronary artery tree from CTA data and make it available to the interventional cardiologist during a procedure. A CT-based, 3-D tree potentially provides an accurate pathway to the lesion, length, vessel size, tortuosity, and angulation of a lesion. Choosing a guide catheter, guidewire, balloon, and a stent will be part of the CT-based preplanning before the patient even enters the catheterization laboratory. This can also be easily applied to complex structural heart disease evaluations like ASD, PFO, pulmonary veins for ablation procedures, and transcatheter interventions of heart valves. This preplanning will likely affect procedural outcomes and make interventional procedures safer and more efficient.

A more advanced use of CT-derived 3-D images of coronary arteries and cardiac chambers is their in-room use to guide interventions. In the coronary arteries, the CT/3-D tree has the ability to deliver an optimal view map (OVM) that identifies the least amount of foreshortening and overlap and, therefore, provides in-room guidance of the optimal projection view. Currently, the operator generated optimal view is based on a “trial-and-error” technique; a CT-based 3-D optimal view provides computer assistance that will likely minimize contrast and fluoroscopy during interventions. The need for image guidance of some structural heart disease interventions is often greater than in the corony arteries because of the need to navigate in 3-D space without the confines of a vascular tree. Operators have been forced to use image fusion with echocardiography to guide procedures, due to the obvious disadvantages of current X-ray technology. While CTA provides a thorough evaluation of the coronary arteries and heart structures, it cannot be presently used online in the guidance of percutaneous coronary interventions (PCI) or in structural heart disease interventions. An image fusion concept that incorporates X-ray and CT should improve the limitations of each technology incorporating the best of both worlds. While this image integration has emerged as an offline evaluation, we have been able to interactively load the CT data onto the X-ray live image to successfully guide interventional procedures.

Selective Coronary Angiography Evaluation

Invasive coronary angiography provides a 2-D representation of a patient’s 3-D coronary anatomy. There are well known limitations of this technique, with vessel foreshortening, overlap and unappreciated angulation/tortuosity being the most common.^5–7 Despite the acquisition of multiple angiographic views in an effort to overcome these limitations, quantitative measurements of vessel properties such as length, diameter and orientation remain limited by foreshortening of the vessel segments and by unknown magnification factors.⁴ Additionally, in the assessment of percent lesion stenosis, despite the use of automated vessel detection tools and calibrated quantitative coronary angiography on 2-D images, significant inaccuracies due to the frequent eccentric nature of coronary artery stenoses remain.^4,8 These limitations become even more important when coronary interventions rely on imaging for safe and efficient evaluations that will lead to procedural success.

Structural Heart Disease Evaluation

Traditionally, echocardiography has had a pivotal role in the evaluation of structural heart disease like PFO, ASD, pulmonary veins, mitral valve disease, and aortic valve disease in the cardiac catheterization laboratory. X-ray by itself has several limitations when it comes to PFO/ASD closure, mitral valvuloplasty, pulmonary vein ablation, and aortic valvuloplasty. For this reason, echocardiographic image guidance in the catheterization laboratory has played up until now a lead role during these interventions. These include transthoracic echocardiography (TTE), transesophageal echocardiography (TEE), and intracardiac echocardiography (ICE). A more comprehensive image fusion concept that streamlines these procedures is absolutely necessary, as they are time-dependent and require additional personnel and/or the use of more expensive equipment.

CTA and 3-D Reconstructions

Currently, there is much interest in the assessment of coronary artery disease (CAD) using multidetector computed tomography (MDCT). Recent advances in MDCT have provided the opportunity to noninvasively and three-dimensionally evaluate the coronary vasculature and heart structures in a safe and efficient manner. Newer CT imaging technologies with faster gantry rotations, dual X-ray source scanners, multidetector 64-row acquisitions and ECG gating has substantially improved both temporal and spatial resolutions to adequately visualize the moving coronary vasculature. While this continues to improve, current-generation MDCT scanners are able to achieve a spatial resolution of 0.4 mm with a temporal resolution as low as 83 msec during cardiac acquisition of less than 15 seconds. Initial, relatively small studies evaluating the diagnostic accuracy of 64-slice-MDCT compared with diagnostic cardiac catheterization have demonstrated sensitivities ranging from 80–94% and specificities ranging from 95–97%.^9–17 Routine evaluation of coronary MDCT involves segmentation of the individual visualized coronary vessels. From the resulting coronary tree, determinations are easily made regarding vessel length, curvature, branching angles, stenosis length, location and severity, which are very useful in the catheterization laboratory. Additionally, atherosclerotic plaque composition can be easily assessed. Due to high CT attenuation of calcified lesions, they are differentiated from fibrous or lipid-rich lesions. These angiographic features are easily displayed on MDCT-derived 3-D volumetric and anatomic representations.^18,19 Three-dimensional reconstruction images are the backbone of the CT/X-ray fusion concept and can be generated from different imaging modalities such as rotational angiography, computed tomography, or magnetic resonance imaging (MR). Several methods that are capable of generating 3-D images have been described and in general, can be divided into surface-rendering or volume-rendering techniques. The surface rendering method relies upon a computer algorithm to reconstruct intensity values, which are above a specific defined threshold and represent volumetric surfaces within the dataset; all values below the set threshold are discarded and not used for image generation.²⁰ The resultant image is a representation of the surface contour and visually appears three-dimensional through computer-generated shading. While surface rendering is fast, using only a small portion of the acquired data, it is less reliable for structures smaller than 2–3 mm.²⁰ The maximum intensity projection (MIP) algorithm is another commonly used volume-rendering technique.²¹ Three-dimensional imaging using volume rendering is a more powerful technique that incorporates the entire dataset into the 3-D image. In contrast to surface rendering techniques, intravascular details and spatial relationships between adjacent structures are preserved.

Optimal View Map Generation

Three-D vascular trees generated with CTA can be used to simulate all possible angiographic views of the vascular tree in the form of an optimal view map. These 3-D datasets can be used to simulate 2-D images in a similar format to those currently employed for guidance of endovascular interventions such as fluoroscopy and ultrasound. With the expanded use of CT, it will be increasingly important to maximize the use of the information from the diagnostic modality when the patient comes for an interventional therapy (coronary or structural). The clinical value of using a 3-D vascular tree to simulate angiographic views is to enhance patient safety and potentially improve interventional outcomes. Computer selection of an optimal view can be done prior to the intervention as part of the preprocedure planning process. The placement of the gantry in a location to produce useful angiographic information is a fundamental task in both diagnostic imaging, but also in the performance of endovascular interventions. Obtaining optimal angiographic views is critical to assessing lesion morphology, extent of disease, and involvement of major branch segments.⁸ These considerations have become more prominent since the advent of interventional cardiology, as the objective has become significantly more demanding than simply noting the quality of distal conduits for bypass surgery. In essence, 3-D vascular trees registered or aligned in the coordinate system of the gantry location can be used to solve the imaging tasks commonly encountered. This feature is now available for the coronary vasculature, but it has not been developed for structural heart disease. Several methods can be used to produce useful images that avoid overlap and minimize foreshortening for all segments of interest in the coronary tree. Computer graphics can be used to display the tree in a variety of views and the operator can select appropriate views. Alternatively, algorithms can be written to automatically process the data, recommend specific views, or produce visual guides that combine a parameter, such as the extent of foreshortening for a vessel segment of interest for all angiographic views.

The Multimodality Fusion Concept

Multimodality fusion. In theory, any type of medical image can be fused or combined with any other type to create a third image. As an example, in contemporary clinical practice, PET or SPECT images are most often fused with CT or MR scans. A first step in this integration process is to bring the modalities involved into spatial alignment, i.e., mapping each image to a common coordinate system, a procedure referred to as “registration”. After registration, a fusion step is required to visualize the integrated information from the data involved based on the assignment of different colors and degrees of opacity or transparency on every pixel or voxel of the integrated datasets. By use of internal landmarks such as anatomical structures or external markers, the process of registration can be performed by (1) initially pure manual operation; (2) manual alignment followed by the computer-assisted process; or (3) automatic computer-based process. The first technique is highly operator-dependent and requires considerable skill. The pure computer-based algorithms can also be a time-consuming approach. Although hybrid systems produce images that are intrinsically registered because the scans are obtained with the patient in the same position, it still requires an alignment process to improve the registration error due to cardiac or respiratory motion for cardiac or abdomen imaging.^22–24 Navigation. Typically, a navigation system consists of a 3-D tracking system and a 3-D multimodality imaging system. The 3-D tracking system generally consists of two major components (a) a sensor unit that can be an optical, magnetic, radio signal, or a microwave-based control device; and (b) an emitter unit that can emit signals such as infraray, magnetic field, radio frequency or microwave. The 3-D multimodality imaging system generates visual information based on available imaging modalities such as anatomical-based CT or MR images, or functional based PET or SPECT data.^25,26 Prior to initiating the tracking process, the 3-D multimodality-based dataset needs to be registered with the patient’s location and orientation. An X-ray angiography-based 3-D reconstruction can be employed to generate desired anatomical landmarks that appear in the 3-D multimodality dataset to facilitate the registration process. In an endoscopic or intervention-based navigation,^27–29 the emitter unit is located at the tip of a catheter or an endoscopic probe. That location of the catheter tip must be matched and displayed within the patient’s anatomy using a previously gathered 3-D multimodality dataset. Navigation systems linked to interventions must have accurate, perfectly registered, and motion-compensated datasets if they are truly to be able to replace the current continuous fluoroscopic image guidance. The movement of vascular trees must be incorporated into 3-D datasets used by navigation systems. The continuous motion of the coronary tree can be incorporated into a 4-D dataset, but then must be combined with sensing a catheter that is also moving due to the tree motion. Navigation systems for guidewire advancement are potentially more readily achievable than systems that position stents due to the risks of malregistration. Vascular datasets with a high degree of temporal resolution are currently not available from CTA and MRA, but only 3-D data derived from X-ray imaging may be superior.

Offline 3-D X-ray/CTA Imaging to Assist Interventions

Coronary interventions. By providing an accurate assessment of coronary lesion length and reference vessel diameter, 3-D datasets of the coronary anatomy allow objective decisions regarding the length and diameter of balloons and stents used to treat obstructive disease.4 The accuracy of such assessments should reduce the incidence of events such as oversizing of balloons, over- and undersizing of stents, and inadequate lesion coverage by drug-eluting stents, resulting in placement of additional stents. Such events are clinically relevant, impacting the safety, efficacy and cost of interventional procedures. The ability of 3-D datasets to assess the orientation of coronary artery ostia, and to measure vessel tortuosity and calcification, together with comprehensive lesion assessment, should also greatly facilitate the choice of guide catheter and guidewire for a given intervention. Currently, the evaluation of these elements results in educated judgments regarding the particular guide shape and caliber required to provide support to deliver interventional equipment, and the specific type of wire that will provide sufficient support for device delivery. High degrees of spatial and temporal resolution are needed to visualize the fine structures of these devices that are implanted in dynamic structures, such as the coronary arteries. At this time, X-ray imaging offers a variety of advantages of providing high-quality images without artifacts. Reconstruction algorithms using rotational angiographic or fluoroscopic acquisition that are currently in development for vascular structures can be applied to implanted devices.30 When successfully applied to coronary stents, this 3-D approach could replace the need for intravascular ultrasound. Finally, the use of rotational fluoroscopy, 3-D reconstruction, and advanced analysis tools will allow a means to monitor the state of implantable devices with a simple, low-radiation, and noninvasive examination.

Structural heart disease interventions. Current technology only allows for the use of image integration in the catheterization laboratory in the form of echocardiography and to a lesser extent, MR in selected locations. Traditionally, echocardiography (TTE, TEE, ICE) has served as the basis of image integration in the catheterization laboratory when it comes to the treatment of ASD, PFO, mitral valve, aortic valve and pulmonary vein isolation in interventional and electrophysiology procedures. Current CT technology advances have shown the potential use of this technology in evaluating defect sizes, orientation, location, and surrounding structures, leading to more efficient procedures and improved outcomes. While the validation of this CT-based evaluation is well underway, the use of CT has also allowed for the creation of physical models that give the individual operator a complete perspective of the heart structures. In the future, this model creation may even be part of a preprocedural assessment and device deployment practice that could likely improve success in difficult cases.

Online X-ray/CTA Fusion-Based Interventions

Image acquisition and overlay registration process. A contrast-enhanced MDCT was performed using a 40-slice scanner (Brilliance 40, Philips Medical Systems, Cleveland, Ohio) using retrospective ECG gating. Scanning was performed at 120 kVp, with an effective tube current of 800 mAs, slice collimation of 40 x 0.625 mm, gantry rotation time of 0.4 seconds and a pitch of 0.2. Images were reconstructed at 0.8 mm slice thickness at increments of 0.4 mm. The images were reconstructed using CB kernel (i.e., standard reconstruction kernel). An ECG-triggered dose modulated protocol (DoseRight, Philips Medical Systems) was used to decrease radiation dosage by cutting down on the tube output during the systolic phases of the cardiac cycle. The resultant radiation dose was ~8 mSv. The total acquisition time taken to cover the cardiac anatomy was 10 seconds. The contrast enhancement was obtained by using a total of 80 ml of contrast medium (Ultravist 370 mg I/ml, Schering AG, Germany) injected into the antecubital vein at a flow rate of 5 ml/s followed by a 50-ml saline chaser bolus. The reconstructed images are then transferred to a dedicated CT workstation (Extended Brilliance Workspace, Philips Medical Systems) and loaded into a dedicated cardiac application (Comprehensive Cardiac Analysis). Automatic segmentation of the cardiac chambers and coronary arteries is performed. After the semi-automatic extraction of the coronary arteries, the areas of interest can be interactively analyzed using multiplanar reformats (MPRs) and 3-D rendering tools. Once the decision is made to send the patient to the cath lab, the CT data and the extracted coronary arteries are sent over using a DICOM push to the three-dimensional imaging (3-DRA) workstation that is positioned in the control room of the cath-lab. At the time when images from the same patient are acquired in the cath lab, a registration process is needed in order to align the pre-operative CT dataset with the images as acquired by the X-ray imaging system. The general approach is to have the operator perform fluoroscopic imaging in the straight AP positionand one at the lateral position. Then, based on the bony landmarks (spine, ribs, wires) visible in both datasets, the two are aligned. Once the pre-procedure CT is aligned with the X-ray imaging system, the operator can freely use the moveable gantry system without having to worry about “losing” the registration. Changes in the X-ray equipment (such as different detector size, SID, viewing angle) are directly relayed back to the 3-DRA workstation, and corresponding adjustments are made to the registration. If, for example, the operator moves the C-arm to a traditionally used viewing angle to visualize the proximal LAD, the corresponding CT visualization will already give a very close idea of how this view may look even before any acquisition is made with the X-ray equipment. Moreover, if it turns out that the specific vessel segment of interest is severely foreshortened or overlapped by some other branches, an alternative viewing angle can be found, perhaps with the aid of the interactive “TrueView” map. Similar to the functionality on the CT workstation, different multiplanar reformats are present on the 3-DRA workstation. This allows the operator to find a particular viewing angle, not only with respect to minimum foreshortening and overlap, but also the luminology. The ability to choose views that optimally visualize the characteristics of the lesion is especially powerful in the presence of an asymmetric stenosis. Apart from integrated visualization of the 3-D CT dataset, the 3-DRA workstation allows for 3-D roadmapping functionality. In this application, the images acquired with the X-ray system (be it fluoro or cine runs) are sent in real time to the workstation where it is shown together with the perspective rendered CT dataset. Though in its present stage of development the CT data are fixed while the angiograms are showing cardiac and respiratory movement, we feel that having these two modalities visualized together on one screen increases the level of confidence of the operator while performing the intervention.

Coronary interventions. The offline evaluation of CT data allows for preprocedural planning. The active overlay of a coronary tree to X-ray angiography should provide the basis for accurate image guidance (ostium cannulation) and lesion location for stent positioning and deployment. The size of the reference vessel segment is already part of the preprocedural evaluation. The current limitation for the use of on-line interactive X-ray/CT overlay is the fact that the coronary tree is moving (this includes rotation, flexion, angulation, torsion, and respiratory motion), making it difficult to fit a static 3-D reconstructed model to the X-ray-based angiogram. Future technology should allow for better interaction of both imaging technologies by having gated, reconstructed images fit the real-time angiograms better. In the meantime, this technology allows for a safer and more efficient way to perform coronary interventional procedures with the existing CT data. The development, use, and validation of the interactive overlay are works in progress.

Structural heart disease interventions. As opposed to coronary interventions, structural heart disease procedures are exposed to a lesser extent to conformational and positional changes of the anatomy. While respiratory motion, systole, and diastole should be taken into account, they have little impact on defects like ASD and PFO. The interactive X-ray/CT overlay allows for navigation of the catheter to the septum, crossing of the defect, balloon sizing, and device deployment. Our group has successfully guided two cases of ASD closure with the X-ray/CT overlay. Its use has decreased procedural time and the use of intraprocedural ICE, and will likely increase procedural success. While the use of X-ray/CT overlay has now become feasible, studies must be performed to validate and evaluate its safety and procedural success.

Conclusion

Standard angiography of the coronary vasculature and heart structures is limited by its 2-D projection of complex 3-D structures and the consequent imaging artifacts that limit interpretation and analysis. Despite the combined modalities of X-ray and echocardiography, percutaneous treatment of intracardiac shunts and valvular disease is procedurally cumbersome and fraught with some difficulties in device selection and alignment of the device to the structures. Complicated percutaneous procedures require precise imaging guidance, and conventional X-ray is often limited in its utility. By combining X-ray imaging with real-time, interactive, anatomical, CT-based datasets, interventional procedures could be facilitated. Our group has been able to transfer the extracted CT data to the catheterization laboratory environment by virtue of new commercially available 3-D software. CT-based 3-D reconstruction provides the accurate length, size, tortuosity, and angulation of a lesion, thereby improving procedural success in our pilot evaluations. Moreover, the system can clearly identify an optimal working view where foreshortening and overlap are minimal, without the use of additional contrast or fluoroscopy. CT acquisition quality, however, is dependent on low resting heart rates (50–60 beats per minute), thus possibly limiting its applicability to a broad range of patients. Despite its potential benefits, use of the current CT technology has constraints. Even while reducing procedural contrast and radiation exposure in the catheterization laboratory by providing preprocedural information, current CT techniques still use high doses of both. It is important to note that not all patients will need a CTA before going to the catheterization laboratory. As CTA matures, the indications and guidelines will be more detailed and explicit as to the patient population that benefits the most from it. Those individuals with a prior CT evaluation have the ideal platform for image integration with X-ray, and the data use should be maximized.

The perfect imaging technique should be safe, fast, inexpensive, preferably noninvasive, and reproducible. In coronary angiography and percutaneous interventions, this translates into minimization of fluoroscopy and contrast, safer imaging techniques, unforeshortened and nonoverlapped views, and a more efficient procedure. This is exactly what merging CTA results and 3-D reconstructions will provide to the intervention arena. Additionally, with this image reconstruction software tangible, 3-D models may be created and utilized to improve device selection, sizing, physician training and may even advance device design in structural heart disease. Multimodality integration will ultimately advance current imaging techniques and improve patient care.