Multidetector-row Computed Tomography in the Evaluation of Heart Failure

Login or register to view PDF.
Abstract

Cardiac computed tomography (CCT) has undergone significant improvements in recent years with the development of multidetector-row and dualsource scanners. This has led to a remarkable improvement in CCT’s temporal and spatial resolution and better capability to visualize various cardiac structures, including coronary arteries. In addition to evaluation of the coronary tree, CCT allows evaluation of multiple other cardiac anatomical and functional parameters, including estimation of bi-ventricular volumes, dimensions, regional wall motion, diastolic function, and ejection fraction, all without additional imaging or contrast administration. CCT also allows evaluation of pulmonary venous as well as cardiac venous anatomy. Moreover, studies have been conducted on the utility of CCT to detect synchronization of myocardium after bi-ventricular lead placement. Furthermore, CCT perfusion imaging, pre-operative assessment of patients prior to valve replacement, and routine follow-up of transplant patients for atherosclerosis are under active investigation. However, limitations include radiation exposure, need for contrast administration, and inability to provide physiological as well as hemodynamic information across valves. The aim of this review is to summarize the current state of evidence for multidetector-row CT (MDCT) applications for patients with proven or new-onset congestive heart failure.

Disclosure
The authors have no conflicts of interest to declare.
Correspondence
Matthew J Budoff, MD, FACC, FAHA, Los Angeles Biomedical Research Institute at Harbor-UCLA, Torrance, CA 90502. E: mbudoff@labiomed.org
Received date
06 January 2010
Accepted date
11 February 2010

The prevalence of heart failure and the resultant mortality has continued to rise despite increased understanding of the pathogenesis and improvement in management strategies.1 Left ventricular (LV) dysfunction is the final stage of most primary cardiovascular diseases, and greater severity is directly correlated with worse prognosis. The principal manifestation of heart failure progression is a change in the geometry and structure of the LV, such that the chamber dilates or hypertrophies and becomes more spherical, termed ‘cardiac remodeling.’2 This may occur through either direct loss of myocardial tissue (necrosis or apoptosis) or altered hemodynamic process secondary to progression of coronary artery disease (CAD), diabetes, hypertension (HTN), or atrial fibrillation (AF). LV remodeling can progress even in the absence of clinical signs and symptoms of heart failure.3

The information provided by imaging techniques is used to guide clinicians to intervene early and undertake appropriate preventive and therapeutic measures. Non-invasive techniques can identify patients with systolic as well as diastolic LV dysfunction, and can also differentiate dysfunctional but viable myocardium from non-viable myocardium in case of ischemic cardiomyopathy (CMP). Measurements of LV mass, volume, and systolic function are accepted as end-points in both clinical practice and research.4

Recent advances in cardiac multidetector-row computed tomography (MDCT) technology now include higher spatial and temporal resolution, permitting contrast-enhanced imaging of coronary arteries during a single breath-hold. In addition to the increasing number of studies validating obstructive coronary disease detected on MDCT with conventional coronary angiography, multiple other applications involving non-coronary structures including both anatomical and functional parameters have been studied, all of which are highly applicable to patients with heart failure. This review will elaborate on the potential applications of cardiac MDCT in the diagnosis and management of such patients.

Use of Multidetector-row Computed Tomography to Determine Etiology of Congestive Heart Failure (Dilated versus Ischemic Cardiomyopathy)

In a patient who presents with new-onset heart failure, determination of the cause of CMP is crucial for management (see Figure 1). Perfusion defects have been found to exist in both ischemic and non-ischemic etiologies, and physiological assessment has not been demonstrated to be very accurate in this setting.5 Anatomical approaches, such as evaluation of coronary artery calcification (CAC) by electron-beam computed tomography (EBT), have been used to differentiate ischemic from dilated CMP.6–9 The study by Budoff et al.9 compared nuclear stress imaging (NSI) versus EBT in 56 patients with new-onset CMP who subsequently underwent invasive cardiac catheterization (ICA). NSI was found to have a sensitivity and specificity of 97 and 18%, respectively, while an EBT CAC score of >0 Angstrom units (AU) had a sensitivity and specificity of 97 and 68%, respectively, for discriminating ischemic from non-ischemic CMP. The overall diagnostic accuracy of EBT was 84% versus only 64% with nuclear imaging (p=0.009).

Danciu et al.,10 in their study on 421 symptomatic patients with intermediate risk for CAD, compared MDCT versus NSI. One of the criteria for an abnormal NSI was normal or <10% perfusion defect in the presence of newly detected LV dysfunction (<35% ejection fraction [EF]) at rest. After NSI and MDCT assessment, 18.5% were sent for ICA and 81.5% were medically managed; all patients were followed up at an average of 15 months. In the group referred for ICA, 50 cases required immediate revascularization, one experienced non-ST-segment-elevation myocardial infarction, there was one death, and five patients required repeat ICA, three of whom underwent late revascularization. In the medically managed group, six patients required late ICA, one of whom underwent revascularization. This study clearly indicates that MDCT can identify up to 80% of patients at low to intermediate risk of events, including patients with new-onset LV dysfunction where invasive cardiac catheterization can be safely avoided. Andreini et al.11 performed 16-slice MDCT and ICA in 61 patients with dilated CMP of unknown origin and 139 patients with normal EF. In patients with dilated CMP, the overall feasibility of coronary artery visualization was 97.2% versus 96.1% in the control group, and all cases except one with normal or pathological coronary arteries by ICA were correctly detected. None versus 10 complications occurred with MDCT versus ICA (p=0.001).

The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of MDCT for the identification of >50% stenosis were 99, 96.2, 81.2, and 99.8%, respectively, compared with 86.1, 99, 96.4, and 99.8%, respectively, for the control group. Cornily et al.12 similarly evaluated 36 patients with dilated CMP who underwent MDCT before ICA with ventriculography. On patients with Agaston score equivalent (ASE) <1,000, the sensitivity, specificity, PPV, and NPV of MDCT in detecting significant CAD was 100, 80, 67 and 100%, respectively, thus avoiding ICA successfully in 77.7% of the patients. For ASE ≥1,000, MDCT enabled ICA to be avoided in only 22.2%. Thus, when ASE is <1,000 and MDCT reveals <50% stenosis of coronary arteries, this technique may be used solely to exclude CAD and may potentially replace ICA. Given the huge potential of cardiac CT to discern ischemic from non-ischemic CMP more safely and less expensively than either NSI or ICA, further studies are needed to consider MDCT as the chosen diagnostic modality to evaluate new-onset CMP.

Use of Multidetector-row Computed Tomography to Assess Perfusion and Viability

Perfusion imaging by MDCT is being actively pursued with cardiac CT. Correlating the stenotic plaque in the coronary artery with the perfusion defect simultaneously by using the same imaging modality further strengthens the diagnosis of ischemia in a patient with CMP. Viability imaging by MDCT is based on the ‘delayed enhancement’ (DE) phenomenon, as in magnetic resonance imaging (MRI) to detect myocardial scar versus viability,13 while myocardial density analysis via ‘perfusion’ is used to visualize areas of low blood flow in the myocardium.

In an acute phase of coronary occlusion with viable myocardium, the revascularization can significantly improve ventricular dysfunction; by contrast, the presence of non-viable myocardium increases exposure to the unnecessary risk of invasive procedures and increases late mortality. Single-photon-emission computed tomography (SPECT) plays an important role in the identification of tissue viability in myocardial segments. Metabolic imaging with positron-emission tomography (PET) offers regional tissue viability in patients with advanced CAD and severely impaired LV function. DE magnetic resonance angiography (MRA-DE) has been well validated over the past several years and is performed routinely by several clinical centers. However, as the clinical indications for implantable cardiac defibrillators and bi-ventricular pacing continue to expand, and with inability to perform MRI on these patients, newer imaging modalities such as MDCT can be used to acquire this important information before pursuing invasive cardiac catheterization.

The potential role of CT in the detection of acute myocardial infarction in explanted hearts and experimental animal models was first noted in the late 1970s by Higgins et al.14–18 In this series of work, it was first shown in explanted hearts that acute infarcts were detectable by cardiac CT as regions of hypo-enhancement compared with the normal myocardium.14 Higgins et al.19,20 performed fluorescent excitation analysis to demonstrate preferential uptake of contrast media in myocardial segments damaged by acute ischemia. Hoffmann et al.21 performed four-slice MDCT in a pig model of non-reperfused myocardial infarction and showed a good correlation between triphenyltetrazolium chloride-derived infarct region and MDCT regions of hypo-enhancement.

Manken et al.22 compared MRA-DE with conventional MDCT images and post-processed images with color coding based on attenuation values. They found an agreement between conventional MDCT and MRA (κ=0.756) and MRA with post-processed images (κ=0.850). Jacquier et al.23 performed MDCT on 19 patients and found that imaging performed five minutes after injection yields a higher signal-to-noise ratio and image quality than imaging performed after 10 minutes, with no difference in the extent of infarct measurement. Similarly, Habis et al.,24 immediately after ICA without iodine re-injection, performed MDCT and then MRA after 10±4 days. On segment-based and patient-based analysis, sensitivity, specificity, accuracy, PPV, and NPV of MDCT versus MRA were, respectively, 84, 96, 94, 85, and 96% versus 90, 80, 88, 95, and 67%. Myocardial infarct size assessed by the two methods was highly correlated (r=0.94; p<0.0001). Le Polain et al.25 specifically looked at 71 patients with CMP of unknown cause. MDCT coronary assessment and DE was compared with ICA findings of significant stenosis (>50%) and MRA-DE. ‘Definite ischemic CMP’ was defined as ICA showing >50% stenosis and evidence of transmural or subendocardial DE by MRA (24 patients), while ‘non-ischemic CMP’ was defined as <50% stenosis on ICA and no or atypical DE on MRA (36 patients). Combined coronary and DE-MDCT was found to have an excellent agreement (κ= 0.89; p<0.001) with ICA/MRA-DE to classify patients into the above categories on perpatient- based analysis. Sensitivity, specificity, and accuracy of MDCT were 97, 92, and 94%, respectively, for detecting patients with definite ischemic LV dysfunction (see Figure 2).

The 0.5mm slice thickness of MDCT technology is 10–20 times thinner than that used in typical MRA viability imaging studies, thereby greatly reducing partial volume effects and greatly improving spatial resolution in the z axis, at the expense of temporal resolution. A 0.5mm MDCT slice achieves near isotropic resolution, with the ability to reconstruct any arbitrary slice orientation from the original stack of axial slices. This is particularly practical for viability imaging, where infarct morphology can vary substantially when slices are oriented even slightly obliquely from a true short-axis view.

Further studies with reproducible results are required in this area to recommend the use of MDCT-DE in combination with anatomical finding of coronary disease as a means to rule out ischemia as the etiology behind CMP, as well as to determine the viability of the involved segment to assess the potential benefit of revascularization. Significant validation is available from MRA, while CT data are forthcoming.

Use of Multidetector-row Computed Tomography to Measure Ejection Fraction and Ventricular Volumes to Monitor Therapy and Determine Systolic versus Diastolic Dysfunction
Left Ventricular Structure, Function, and Chamber Size

Assessment of LV function including stroke volume and ejection fraction is now possible in a study where images are acquired via retrospective electrocardiogram (ECG) gating (see Figures 3–5). Currently, MRA is considered the gold standard for ventricular assessment. However, due to the higher spatial resolution of MDCT compared with MRI, it is at least equivalent or most likely superior with regard to volume and mass measurements.26 There are certain situations where MRA is contraindicated, e.g pacemaker/implantable cardioverter–defibrillator (ICD) implantation or when a patient with CMP cannot hold his or her breath for long periods of time. MDCT can be used as the preferred non-invasive diagnostic modality in such patients; however, the concern of higher radiation exposure with retrospective ECG gating remains. Newer radiation-saving modalities could reduce the radiation exposure, but with the concurrent medical conditions of some patients—including diabetes and presence of impaired kidney function—contrast-induced nephropathy can be a risk and requires precautious use of contrast. An advantage of MDCT for accurate assessment of LV systolic function is the fact that it measures LV cavity size without geometric assumptions.

Several studies have shown excellent correlation between MDCT and other imaging modalities for LV functional assessment27–32 of 0.80–0.98 for LVEF assessment. In a meta-analysis of studies comparing assessment of LVEF by MDCT and MRI,33 it was noted that when LV function parameters were grouped by increasing number of detector rows, there was an increase in the diagnostic performance of MDCT.

Whether this correlation holds true in patients with lower EF is a question asked by few studies. Ramon et al.34 assessed bi-ventricular size and systolic function via MRA and MDCT in patients over a wide range of ejection fractions (30–72%) and found a very high correlation for end-systolic volume (ESV) and end-diastolic volume (EDV), as well as mass. In 18 cardiac magnetic resonance (CMR)–CT data pairs, right ventricular (RV) EF showed moderate agreement (r=0.86), and RV volumes correlated well (r=0.97 and 0.94 for RV EDV and RV ESV, respectively). For segments adequately visualized by both techniques, the mean κ statistic was 0.88 (range 0.78–1.0), consistent with good agreement for wall motion assessment. Butler et al.35 studied patients with lower EF (mean EF by echocardiography 36±8% versus 38±12% by MDCT; r=0.67, p=NS). In this study the mean LV end-diastolic and end-systolic diameter by echocardiography and MDCT were 56±8 and 46±9mm and 58±12 and 47±11mm, respectively (r=0.71 and 0.77, respectively; p>0.20 for both). Mean lateral and septal wall thickness by echocardiography and MDCT were 10±1.4 and 11±1.5mm and 10±1.3 and 10±1.4mm, respectively (r=0.77 and 0.76, respectively; p>0.20 for both). Mean LV EDV, ESV, and stroke volume by echocardiography and by MDCT were, respectively, 123±45, 78±31, and 44±21ml and 140±58, 92±43, and 48±24ml (r=0.62, 0.67, and 0.60; p>0.20 for each). The regional wall motion assessment correlation was good between the two modalities (κ=0.61). Inter-observer correlation between the two MDCT readers ranged from good (r=0.72 for LV EDV) to excellent (r=0.84 for septal wall thickness).

Myocardial Mass Assessment by Multidetector-row Computed Tomography

Myocardial mass is measured in a very similar manner to ventricular volume (see Figures 3 and 4). Most post-processing tools offer the ability to draw an epicardial border as well as an endocardial border. The difference between both volumes represents the myocardial volume of the left ventricle.

The myocardial mass is then calculated by multiplying this volume by the voluminal weight for myocardial tissue. In the meta-analysis by Yamumuro et al.,36 in seven studies data concerning the myocardial mass of 10 cohorts were provided. The average myocardial mass measured by MRA was 141.5±24.2g versus 140.8±25.1g by MDCT. An excellent agreement was found between both modalities. Several other studies comparing myocardial mass assessment via various imaging modalities including MDCT and SPECT have recently been published.37,38

Detection of Dyssynchrony

While processing images acquired during a retrospective MDCT scan, it is possible to construct images in both short- and long-axis views, and the cardiac contraction and relaxation cycle can be reconstructed at every 5% of the RR interval. Therefore, segmental wall motion and systolic contraction timing can be assessed in a similar manner to the 3D echocardiographic systolic dyssynchrony index (DI) in patients assessed for bi-ventricular pacemaker implantation.39 However, the temporal resolution of the 64-row MDCT is still inferior to that of tissue Doppler echocardiography.40 Truong et al.41 studied 38 CMP patients with wide QRS complexes. Using 64-row MDCT, LV endocardial and epicardial boundaries were delineated from short-axis images reconstructed at 10% phase increments of the cardiac cycle. A global metric was defined using changes in wall thickness as the DI. The DI was the most reproducible metric (inter- and intra-observer intra-class correlation coefficients ≥0.94; p<0.0001) and was used to determine differences between the three groups: wide QRS group (EF 22±8%, QRS 163±28ms), narrow QRS group (EF 26±7%, QRS 96±11ms), and age-matched control subjects (EF 64±5%, QRS 87±9ms). Mean DI was significantly different between the three groups (wide QRS 152±44ms, narrow QRS 121±58ms, and control subjects 65±12ms; p<0.0001) and greater in the wide QRS (p<0.0001) and narrow QRS (p=0.005) groups compared with control subjects. DI had a good correlation with 2D (r=0.65; p=0.012) and 3D (r=0.68; p=0.008) echocardiographic dyssynchrony (see Figure 6). As MDCT technology advances further, future studies will provide better understanding of the use of MDCT in the detection of ventricular dyssynchrony. MDCT utility in dyssynchrony has not yet been validated.

Right Ventricular Structure, Function, and Chamber Size, Right-sided Heart Failure, and Pulmonary Hypertension

RV function parameters are important for patients with CMP and have prognostic implications; however, since RV myocardium is much thinner, the differentiation between the RV myocardial and right atrial border is more difficult to detect compared with the left heart.42 Cardiac MDCT scanning provides adequate visualization of RV wall thickness and function, free of geometric assumptions. Pulmonary artery diameter and RV wall thickness can be used on MDCT as well to assess pulmonary pressures in patients with heart failure. An initial study using a four-detector-row scanner was reported by Kim et al.43 Koch et al.44 aimed to determine RV function from 16-detector-row CT in comparison with MRA on 19 patients. Mean end-diastolic (155.3±54.6ml) and end-systolic (79.1±37ml) RV volumes correlated well with MRA (151.9±53.7ml, r=0.98, and 75.0±36ml, r=0.96, respectively; p<0.001). RV stroke volume (76.2±20.2ml for MPR-CT, 76.9±20.7ml for MRA, r=0.93) showed a good correlation, and RVEF (50.8±8.4% for MPR-CT, 51.9±7.4% for MRI, r=0.74) only a moderate correlation. In this study the resulting differences between CT to MRI were small, with a mean difference of about 4ml for RV ESV and RV EDV. In MDCT, due to partial volume effects and angular sections of this region, the exact determination of RV contours is complicated. Wuest et al.45 used a newer protocol of contrast injection in dual MDCT. In this study on 106 patients, the dual-flow concept enabled a statistically significantly better delineation of the septum and right ventricle in dual-source cardiac CT for both the quantitative and semi-quantitative analyses. MDCT studies are ongoing in this area.

Use of Multidetector-row Computed Tomography for Established Congestive Heart Failure to Determine Adequacy of Veins Before Bi-ventricular Pacemaker Lead Placement

Knowledge of the anatomy of the coronary venous system (CVS) is important for planning cardiac interventions such as cardiac resynchronization therapy for bi-ventricular lead placement. This requires cardiac MDCT imaging to be performed a few seconds later than for coronary arteries to allow adequate filling of the veins, enable better visualization of various branches and their size, and aid decision-making with regard to percutaneous versus surgical LV lead placement.46,47 Jongbloed et al.,46 while studying the CVS, found separate insertion of the coronary sinus and the small cardiac vein in the right atrium (63%) as the most common variant.

Other variations included continuity of the anterior and posterior venous system at the crux cordis and the posterior interventricular vein not communicating with the coronary sinus. The mean distance from the posterior interventricular vein to the posterior vein of the LV was 42.4±18.1mm, from the posterior vein of the LV to the left marginal vein was 39.9±15.6mm, and from the left marginal vein to the anterior interventricular vein was 45.4±15.3mm. The diameter of the coronary sinus ostium was 12.6±3.6mm in the anteroposterior direction and 15.5±4.5mm in the superoinferior direction (see Figure 7).

Considering these significant variations, pre-procedural coronary venous system mapping may improve the outcomes of bi-ventricular pacemaker placement (see Figure 8). A comparison of retrograde coronary sinus angiography (CSA) and ECG-gated MDCT for the visualization of the CVS in 20 patients with congestive heart failure (CHF) was performed by Knackstedt et al.48 Overall, there was a trend for MDCT to detect more vessels, and this technique was found to be suitable for an overview of the coronary venous anatomy. In a comparison study of MDCT, invasive angiography during device placement, and tri-phase tissue synchronization imaging (TSI) by van de Vdire et al. on 21 patients,49 an excellent agreement between MSCT and invasive venography was found. In 12 patients, a match was observed between the area of latest mechanical activation (on TSI) and LV lead position. These patients showed a significant decrease in LV dyssynchrony with acute reduction in LV ESV and improvement in LVEF. Patients with a mismatch between the area of latest activation (detected with 3D echocardiography) and LV lead position (coronary anatomy depicted with MDCT on a 3D LV volume set) remained dyssynchronous without improvement in LV function. This showed that in the absence of a match between the suitable tributory of coronary sinus (CS) and the area of latest mechanical activation, a surgical approach may be preferred. Similary, Wei et al.50 studied 141 patients who underwent 64-slice MSCT and 3D reconstruction and ICA and CS angiography before radiofrequency catheter ablation, and found a very good correlation in identifying coronary sinus and other cardiac veins.

Assessment of Devices in Patients with Refractory Heart Failure

At many centers LV or bi-ventricular assist devices that perform mechanical unloading of the heart are used as a bridge to transplant in heart failure patients. These devices have been shown to lead to myocardial recovery, which occasionally enables patients to undergo successful device removal. However, there are no conclusive data on how to predict sufficient recovery of the heart; also, visualization of devices is difficult due to the metal artifact in MRI or echocardiography. MDCT has been used in several studies as a successful modality to assess cardiac function in such patients. Takeda et al.51 presented their preliminary data on cardiac function assessment in three patients with an LV assist device during the off-pump period. Future studies are required in this area, and the results may affect the follow-up assessment of such patients when looking for myocardial recovery.

Other Applications of Multidetector-row Computed Tomography in Heart Failure Patients

Heart failure is seen in many patients with congenital heart disease who enter adulthood either treated or untreated, such as those with atrial and ventricular septal defects. MDCT can be used to evaluate various cardiac chambers and coronary vessels in such patients.52–54 Some other applications include pulmonary venous mapping prior to atrial fibrillation ablation in patients who develop atrial fibrillation (see Figure 9), and diagnosis of pericardial calcification (see Figure 10).

Limitations of Multidetector-row Computed Tomography in the Evaluation of Heart Failure Patients

Cardiac MDCT provides adequate information on the anatomy of cardiac valves (number of cusps, thickened or calcified valves, flail leaflet);55,56 however, it does not provide information on the flow and gradients across valves. Therefore, echocardiogram and MRI continue to be the preferred methods of assessing cardiac valves. There are no data on the safety of intravenous beta-blocker administration prior to MDCT scanning in heart failure patients specifically. Therefore, heart failure patients should be prepped on an individual basis under close supervision and with frequent blood pressure monitoring. Additionally, sublingual nitroglycerin is typically given immediately before the scan to improve coronary vasodilatation and luminal visualization.

Caution needs to be exercised with the combined use of intravenous beta-blockers and sublingual nitroglycerin simultaneously in heart failure patients who may be hypotensive or may have significant RV dysfunction. Retrospective gating in MDCT allows images to be acquired at 5 or 10% of RR interval and thus cardiac functional information can be obtained, but with the disadvantage of increased radiation dose. Prospective imaging, which allows marked reduction of radiation dose, does not permit measurement of EF. MDCT protocols require the use of iodinated contrast media, which is associated with nephrotoxicity, and patients with renal insufficiency—which is commonly present in heart failure patients—must be carefully evaluated prior to CT procedures. With improved spatial resolution it is now possible to decrease the amount of iodinated contrast necessary to opacify the coronary arteries to 40–80ml. MDCT is still in the developmental phase in terms of performing perfusion imaging, despite the many studies currently being reported. Valve evaluation by MDCT is still under active evaluation.

Conclusion

MDCT has shown promising results with regard to the assessment of heart failure patients. It not only allows differentiation of ischemic from dilated CMP, but also allows evaluation of LV and RV function, volumes, and mass and provides assistance in commonly performed procedures such as bi-ventricular lead placement for synchrony. It may also be useful in providing accurate quantification of chambers and function. Perfusion imaging, viability, and evaluation of dyssynchrony remain promising future uses of CT that still require further validation.

References
  1. Web alert, Curr Heart Fail Rep, 2008;5(4): online.
    Crossref
  2. Sun Y, Cardiovasc Res, 2009;81(3):482–90.
    Crossref | PubMed
  3. Cohn JN, Ferrari R, Sharpe N, J Am Coll Cardiol, 2000;35:569–82.
    Crossref | PubMed
  4. Zoccali C, Benedetto FA, Tripepi G, et al., J Am Soc Nephrol, 2006;17(5):1460–65.
    Crossref | PubMed
  5. Dunn RF, Uren RF, Sadick N, et al., Circulation, 1982;66:804–10.
    Crossref | PubMed
  6. Budoff MJ, Shavelle DM, Lamont DH, et al., J Am Coll Cardiol, 1998;32:1173–8.
    Crossref | PubMed
  7. Budoff MJ, J Am Coll Cardiol, 1999;34:247–8.
  8. Le T, Ko JY, Kim HT, et al., Clin Cardiol, 2000;23:417–20.
    Crossref | PubMed
  9. Budoff MJ, Jacob B, Rasouli ML, et al., J Comput Assist Tomogr, 2005;29:699–703.
    Crossref | PubMed
  10. Danciu SC, Herrera CJ, Stecy PJ, et al., Am J Cardiol, 2007;100(11):1605–8.
    Crossref | PubMed
  11. Andreini D, Pontone G, Pepi M, et al., J Am Coll Cardiol, 2007;49(20):2044–50.
    Crossref | PubMed
  12. Cornily JC, Gilard M, Gal GL, et al., Eur J Radiol, 2007;61:84–90.
    Crossref | PubMed
  13. Kim RJ, Wu E, Rafael A, et al., N Engl J Med, 2000;343:1445–53.
    Crossref | PubMed
  14. Gray WR, Buja LM, Hagler HK, et al., Circulation, 1978;58: 497–504.
    Crossref | PubMed
  15. Higgins CB, Siemers PT, Schmidt W, Newell JD, Circulation, 1979;60:284–91.
    Crossref | PubMed
  16. Higgins CB, Siemers PT, Newell JD, Schmidt W, Invest Radiol, 1980;15:S176–S182.
    Crossref | PubMed
  17. Doherty PW, Lipton MJ, Berninger WH, et al., Circulation, 1981;63:597–606.
    Crossref | PubMed
  18. Huber DJ, Lapray JF, Hessel SJ, AJR Am J Roentgenol, 1981;136:469–73.
    Crossref | PubMed
  19. Higgins CB, Sovak M, Schmidt W, Siemers PT, Invest Radiol, 1978;13:337–9.
    Crossref | PubMed
  20. Higgins CB, Sovak M, Schmidt W, Siemers PT, Am J Cardiol, 1979;43:47–51.
    Crossref | PubMed
  21. Hoffmann U, Millea R, Enzweiler C, et al., Radiology, 2004;231:697–701.
    Crossref | PubMed
  22. Mahnken AH, Lautenschläger S, Fritz D, et al., Int J Cardiovasc Imaging, 2008;24(8):883–90.
    Crossref | PubMed
  23. Jacquier A, Boussel L, Amabile N, et al., Invest Radiol, 2008;43(11):773–81.
    Crossref | PubMed
  24. Habis M, Capderou A, Sigal-Cinqualbre A, et al., Heart, 2009;95(8):624–9.
    Crossref | PubMed
  25. le Polain de Waroux JB, Pouleur AC, et al., Eur Heart J, 2008;29(20):2544–51.
    Crossref | PubMed
  26. Orakzai SH, Orakzai RH, Nasir K, Budoff MJ, J Comput Assist Tomogr, 2006;30(4):555–63.
    Crossref | PubMed
  27. Juergens KU, Grude M, Maintz D, et al., Radiology, 2004;230:403–10.
    Crossref | PubMed
  28. Mahnken AH, Spuentrup E, Niethammer M, et al., Acta Radiol, 2003;44:604–11.
    Crossref | PubMed
  29. Mahnken AH, Koos R, Katoh M, et al., Eur Radiol, 2005;15:714–20.
    Crossref | PubMed
  30. Dewey M, Muller M, Teige F, Hamm B, Eur Radiol, 2006;16:25–31.
    Crossref | PubMed
  31. Grude M, Juergens KU, Wichter T, et al., Invest Radiol, 2003;38:653–61.
    Crossref | PubMed
  32. Heuschmid M, Rothfuss JK, Schroeder S, et al., Eur Radiol, 2006;16:551–9.
    Crossref | PubMed
  33. Vander Vleuten PA, Willems TP, Gotte MJW, et al., Acta Radiol, 2006;47:1049–57.
    Crossref | PubMed
  34. Raman SV, Shah M, McCarthy B, et al., Am Heart J, 2006;151(3):736–44.
    Crossref | PubMed
  35. Butler J, Shapiro MD, Jassal DS, et al., Am J Cardiol, 2007;99(2):247–9. Erratum: Am J Cardiol, 2007;99(11):1622.
    Crossref | PubMed
  36. Budoff MJ, Ahmadi N, Sarraf G, et al., Acad Radiol, 2009;16(6):726–32.
    Crossref | PubMed
  37. Plumhans C, Keil S, Ocklenburg C, et al., Invest Radiol, 2009;44(8):476–82.
    Crossref | PubMed
  38. Okwuosa TM, Hampole CV, Ali J, Williams KA, J Nucl Cardiol, 2009;16(5):775–83.
    Crossref | PubMed
  39. Leclercq C, Kass DA, J Am Coll Cardiol, 2002;39:194–201.
    Crossref | PubMed
  40. Bank AJ. Kelly AS, J Card Failure, 2006;12:154–62.
    Crossref | PubMed
  41. Truong QA, Singh JP, Cannon CP, et al., JACC Cardiovasc Imaging, 2008;1(6):772–81.
    Crossref | PubMed
  42. Grothues F, Moon JC, Bellenger NG, et al., Am Heart J, 2004;147:218–23.
    Crossref | PubMed
  43. Kim T, Kim S, Ryu Y, et al., Eur Radiol, 2004;14(S2):S270.
  44. Koch K, Oellig F, Oberholzer K, et al., Eur Radiol, 2005;15:312–18.
    Crossref | PubMed
  45. Wuest W, Zunker C, Anders K, et al., Eur J Radiol, 2008;68:392–7.
    Crossref | PubMed
  46. Jongbloed MR, Lamb HJ, Bax JJ, et al., J Am Coll Cardiol, 2005;45:749–53.
    Crossref | PubMed
  47. Abbara S, Cury RC, Nieman K, et al., Am J Roentgenol, 2005;185:1001–6.
    Crossref | PubMed
  48. Knackstedt C, Mühlenbruch G, et al., Int J Cardiovasc Imaging, 2008;24(8):783–91.
    Crossref | PubMed
  49. Van de Vdire NR, Ajmonte-Marsan N, Schuijf JD, et al., Am J Cardiol, 2008;101:1023–9.
    Crossref | PubMed
  50. Wei Y, Xie P, Pang W, et al., Int J Cardiol, 2009;137(3):276–81.
    Crossref | PubMed
  51. Takeda K, Matsumiya G, Matsue H, et al., J Thorac Cardiovasc Surg, 2008;136(6):1602–3.
    Crossref | PubMed
  52. Manghat NE, Morgan-Hughes GJ, Marshall AJ, Roobottom CA, Heart, 2005;91:1515–22.
    Crossref | PubMed
  53. Bean MJ, Pannu H, Fishman EK, J Comput Assist Tomogr, 2005;29:721–4
    Crossref | PubMed
  54. Castaner E, Gallardo X, Rimola J, et al., Radiographics, 2006;26:349–71.
    Crossref | PubMed
  55. Gilkeson RC, Markowitz AH, Balgude A, Sachs PB, Am J Roentgenol, 2006;186:350–60.
    Crossref | PubMed
  56. Pannu HK, Jacobs JE, Lai S, Fishman EK, J Comput Assist Tomogr, 2006;30:443–6.
    Crossref | PubMed