Optimizing Coronary CTA Workflow: How We Do It

Coronary CT angiography (CCTA) provides an accurate evaluation of coronary-artery disease and coronary-artery anomalies, and it gives us the ability to evaluate the cardiac chambers, myocardium, and valves. Effective deployment of CCTA service requires optimization of workflow to make this procedure cost effective and practical. After more than six years of experience doing all interpretation in volumetric mode, we applied similar workflow-enhancement principles to optimize our CCTA workflow. Baltimore VA Medical Center was the first completely filmless department when it opened its doors in June 1993. Over the past 15 years of being filmless, extensive work has been done in the department to optimize every aspect of the digital imaging chain, from physician order entry to scheduling, acquisition, interpretation, andresults delivery. Similarly, in 2002, the department made another transition to volumetric interpretation, with every CT examination acquired in thin collimation (less than1 mm) and interpretation done in volumetric/multiplanar mode for every study, irrespective of clinical indication. During this transition, volumetric interpretation tools were also deployed enterprise wide to give physicians access to the same tools that radiologists use for diagnosis. In short, we made volumetric interpretation part of routine workflow for radiologists and clinicians. This transition converted what is commonly referred to as postprocessing and brought it into the interpretation process; instead of relying on technologists to create pretty pictures for clinicians and radiologists, the radiologists took control of the data and began interpreting the data as needed. Because clinicians also had access to the same interpretation tools, they could interact with data in their offices, operating rooms, patient rooms, and even homes. This transition also optimized technologists' workflow, so that now they were able to concentrate more on image acquisition and the patient, rather than on creating reconstructions for others. Data Acquisition The success of CCTA depends on team effort, and that team includes the ordering physician, nursing staff, technologists, and radiologists. Many of the steps that ensure acquisition of a good dataset take place even before scanning begins. Appropriate discussion with the ordering physician regarding the indication for the examination, patient preparation, and patient education are crucial. Screening patients for potential contraindications can save time and prevent patient dissatisfaction. A patient who is aware and educated is more interested and readily participates in making the examination successful. At the Baltimore VA Medical Center, friendly and competent staff set the tone. Patients are brought into a separate room where there is low ambient light and soothing music to relax them. They also are provided with warm blankets to prevent shivering during the examination. Patients are given explicit breathing instructions, and they practice several times before starting the scan. Patients are also made aware of the possible effects of contrast material, including warmth, a metallic taste in the mouth, and pelvic tingling. The ultimate goal is to convey a feeling of reassurance to the patient. The CT scan begins with obtaining a calcium score, scanning from the level of the tracheal bifurcation to the bottom of the heart. The calcium score not only provides independent prognostic information, but also helps in determining whether a patient is a good candidate for CCTA. In a patient with very dense coronary calcifications along the length of the coronary vessels and at multiple branch points, it is extremely difficult to interpret CTA images. Table 1 highlights the protocol currently used in performing 64-slice CCTA at our institution. Just as for coronary calcium scoring, the scan range is from the tracheal bifurcation to the bottom of the heart; however, in patients who have undergone coronary-artery bypass grafting, the range is extended to include the top of the chest, in order to identify the origin of the bypass grafts. Table 1. Sixty-four–slice Coronary CT Angiography Protocol Coronary Calcium Scoring Scanning range: tracheal bifurcation to the bottom of the heart Energy: 120 kVp Effective mA: 190 (with dose modulation enabled) Detector collimation: 1.2 mm Slice thickness: 3 mm with 50% overlap Pitch: 0.2 Kernel: B35f Coronary CT Angiography Scanning range: Tracheal bifurcation to bottom of heart Energy: 120 kVp Effective mA: 900 (with dose modulation enabled) Detector collimation: 0.6 mm Slice thickness: 0.75 mm with 50% overlap Pitch: 0.2 Rotation time: 0.33 s Kernel: B25fIV Contrast Administration Phase 1 test bolus: 20 mL contrast at 5 mL/s Phase 2: 70 mL contrast at 5 mL/s Phase 3: 40 mL saline at 5 mL/s Patient preparation for the 64-slice CT scanner includes instructions to consume nothing by mouth for three hours prior to the study and to avoid caffeine the morning of the scan. Patients are given premedication by the ordering physician, as appropriate, if their baseline heart rates are high. If a patient's baseline heart rate exceeds 70 beats per minute at the time of the examination, IV metoprolol in 5-mg doses is administered, up to a maximum of 25 mg. It is important to use caution when administering beta blockers to patients with asthma, severe aortic stenosis, atrioventricular block, or severe left-ventricular (LV) dysfunction. Immediately before the scan starts, 0.4 mg sublingual nitroglycerin is administered (except to patients who have taken sildenafil citrate within the previous 24 hours for erectile dysfunction, as the interaction between the two medications can precipitate hypotension). Table 2 highlights our patient-preparation protocol. Table 2. Patient Preparation 1. No oral intake for three hours prior to examination (no caffeine) 2. IV catheter, 18 gauge, in right antecubital fossa 3. Beta blockade of metoprolol, IV 5 mg and IV as needed (maximum total: 25 mg) 4. Nitroglycerin, 0.4 mg sublingually, immediately prior to scanning The contrast protocol includes administration of 70 to 90 mL of IV contrast material (iohexol, 350 mg iodine per mL) through an 18-gauge catheter in the right antecubital fossa. The contrast-administration protocol comprises 3 phases. Phase 1, the test bolus, involves injection of approximately 20 mL of contrast material at the rate of 5 mL per second while a series of monitoring scans determines the time to peak aortic root enhancement. The optimal scan delay is equal to the time to peak aortic enhancement plus two seconds to achieve optimal coronary-artery opacification. In phase 2, we inject 70 mL of contrast material at 5 mL per second after the optimal scan delay that was calculated in phase 1. In phase 3, we inject 40 mL of saline at 5 mL per second. The benefits of a saline chaser include greater arterial enhancement, a tighter contrast bolus, and a reduction in streak artifact resulting from contrast in the right side of the heart. A saline chaser also allows a reduction in contrast volume of 15% to 20%, which minimizes both cost and the risk of contrast-induced nephropathy. The saline chaser is just one of several contrast factors that, along with patient factors and scanning technique, affect arterial enhancement. Others include the total amount of iodine administered, the injection rate, and the concentration and viscosity of contrast material. ECG Triggering and Dose Modulation In order to achieve motion-free images of the coronary arteries and aorta, cardiac triggering or gating is used in conjunction with breath holding. At the Baltimore VA, CCTA is performed using retrospective ECG gating. Among the advantages of this technique is the ability to scan in helical mode, which permits the acquisition of a volumetric dataset throughout the cardiac cycle. Data from specific parts of the cardiac cycle are then retrospectively synchronized with the ECG signal for image reconstruction. Additional advantages include better isotropic z-axis resolution and reduced dependence on a regular heart rhythm. Retrospective reconstruction is done using a fraction of the R-R interval or an absolute time point prior to or following the R wave. To evaluate cardiac function and valve motion. images are reconstructed at 10% intervals throughout the cardiac cycle at our institution. With retrospective reconstruction, the dataset can be edited to eliminate irregular beats and achieve motion-free images. A major disadvantage of retrospective ECG gating is an increase in the radiation dose. CCTA can result in a radiation dose equal to, or more than, the dose that results from an interventional cardiac catheterization. Dose modulation can minimize the radiation dose through ECG pulsing, which maintains the nominal tube current (mA) only during diastole, when it is most possible to obtain motion-free images of the coronary arteries. Tube current is then reduced to 20% during systole, which allows an overall reduction in radiation dose of up to 50%, depending on the patient's heart rate. Interpretation Image interpretation at our institution may be unique, compared with what is seen in the community, and is a result of our extensive research and experience in interpretation workflow optimization, workflow modeling, visual perception, and human-computer interaction, along with an understanding of how humans interpret image data and how perception plays a role in our interpretation process. These enabled us to remove postprocessing and bring image data navigation into the interpretation process. After careful review of cardiac CT literature, as well as identification of diagnostic questions that need to be answered during interpretation, a CCTA interpretation workflow was created at our institution. Existing automated tools for chest-wall removal, cardiac extraction, and segmentation of the coronaries and LV were also evaluated for their accuracy and consistency to determine whether they should be incorporated into the workflow. We have learned that humans, from a visual-perception point of view, are able to triangulate the position of an abnormality on 2D images better if they have prior knowledge of the 3D orientation of the object under review. Similarly, if the coronary arteries are initially evaluated in 3D on volume-rendered (VR) projections (Movie 1), the position of an abnormality on 2D images is better understood (Figure 1). Therefore, these VR images are also useful for the evaluation of complex anatomy, including bypass grafts, coronary-artery anomalies, and other coronary-artery abnormalities, such as fistulae. Movie 1. Volume-rendered cine of a normal heart helps in better understanding the orientation of coronary vessels in relation to the heart and other mediastinal structures. Figure 1 Figure 1. An automated anatomic mapping tool can help automated workflow by allowing creation of anatomically based templates and views. Axial oblique maximum- intensity projection image and frontal projection from a volume-rendered 3D reconstruction demonstrate accurate mapping of the right coronary artery and left anterior descending coronary artery. The labeling seen in these images was automatically created by the application without human interaction. Images were captured from Aquarius iNtuition Viewer (Terarecon, Inc, San Mateo, Calif). In addition, VR images are ideal for communicating the findings of CCTA to patients, who often understand and appreciate the severity of their disease better on 3D, colored images of their hearts. Similarly, human observers are better able to comprehend the orientation and relationship of anatomic structures if the anatomic structure is evaluated in the orthogonal planes to the orientation of the structure. This has been known in cardiac imaging in other modalities, where images are acquired in orthogonal planes to the axis of the heart. What questions are we answering during the interpretation of CCTA data? There are two basic questions that need to be answered: Are there any coronary abnormalities, and what is the functional status of the heart?Other abnormalities found during CCTA examination (related to valvular or ancillary findings) are usually not related to the indication for the examination. That means, as part of the interpretation process, that it is important to show the best possible view for coronary-disease evaluation and calculation of the LV ejection fraction (LVEF). For this reason, we choose curved planar reconstruction of the coronary arteries with orthogonal images as the primary display layout when evaluating coronary arteries. Similarly, for evaluation of LV function, cine orthogonal images through the long axis of the LV are important (Movie 2). For estimation of the LVEF, we use automated segmentation tools that segment the LV cavity from the myocardium and calculate LVEF and stroke volume throughout the cardiac cycle. Table 3 describes the interpretation workflow used at our institution. Movie 2 Movie 2. A short-axis cine view is automatically generated as part of a left-ventricular functional evaluation that can transition to aortic-valve evaluation automatically, based on the workflow protocol, without the need for separate postprocessing. Table 3. Ten Step Coronary CT Angiography Interpretation Workflow 1. Automated cardiac extraction 2. Evaluation of the right, left main, left anterior descending, and left circumflex coronary arteries on volume-rendered images through all phases 3. Automated curved planar reconstruction of the right, left main/left anterior descending, and left circumflex coronary arteries and their tributaries 4. Evaluation of each coronary vessel in orthogonal planes and maximum-intensity projection 5. Measurement and documentation of any coronary lesion found 6. Automated left-ventricular ejection fraction estimation 7. Multiplanar cine left-ventricle evaluation to assess function and perfusion abnormalities 8. If needed, cine evaluation of cardiac valves 9. Verification of calcium score estimated during acquisition 10. Review of orthogonal multiplanar-reconstruction images for ancillary findings It is important for the interpretation software being used to be capable of automating the workflow and providing the appropriate orientation as needed, in a stepwise fashion, to make the interpretation process fast, smooth, and consistent. Being able to define an interpretation plane (for example, across the aortic valve) and save it as a predefined plane or template that can be used during review of every case will make the interpretation process extremely efficient, as time spent to achieve the correct orientation can be minimized (Figure 2). Figure 2 Figure 2. Axial maximum-intensity projection and axial oblique multiplanar reconstruction images demonstrate accurate mapping of the aortic and mitral valves by an automated software application, creating a predefined three-chamber view (right) for evaluation of the left-ventricular inflow and outflow tract. Images were captured from Aquarius iNtuition Viewer (Terarecon, Inc, San Mateo, Calif). Most vendors offer automated tools for segmentation of the coronary arteries and estimation of coronary stenosis. The degree of coronary-artery stenosis can be reported either quantitatively or qualitatively. Automated software can, either with user assistance or automatically, identify the point of maximal stenosis and identify reference points of normal caliber superior and inferior to the area of stenosis. These tools then calculate and report the quantitative percent stenosis. If referring physicians prefer qualitative reporting, a mild lesion is generally defined as less than30% stenosis, a mild-to-moderate lesion as 30% to 50% stenosis, a moderate lesion as approximately 50% stenosis, a moderate-to-severe lesion as 50% to 75% stenosis, and a severe lesion as more than 75% stenosis. To achieve reproducibility of the data and decrease variability in reporting, it is important to use quantitative values as a guide when interpreting the severity of a stenotic lesion, so that another cardiologist, radiologist, or clinician will get the same result when analyzing the same data. Conclusion CCTA offers many dvantages to patients and physicians. Not only does CCTA offer a noninvasive, rapid means of visualizing both calcified and noncalcified plaque, but it is also a highly accurate tool for the evaluation of coronary-artery stenosis. In addition, CCTA provides valuable information on the cardiac chambers, myocardium, and cardiac valves. Automated software applications based on workflow-driven interpretation and relying on server-side rendering can enable users to make quick, accurate interpretations and can decrease interpretation time until it is similar to that observed for routine CT interpretation. Current technology allows automation of the processing needed for CCTA interpretation so that postprocessing can become interpretation processing or even preprocessing. We believe that, through optimization of the entire interpretation process and by moving to routine volumetric interpretation of all cross-sectional studies, interpretation time can be decreased while maintaining or even improving diagnostic accuracy.