Three-dimensional echocardiography and magnetic resonance imaging allow the volumetric analysis of ventricular volumes independent of geometric assumptions. The aim of the study was to compare these methods and the common angiocardiography in a cardiac model of known volume.
Methods/materials: Right and left ventricular (RV, LV-) volumes were measured in a specific animal model directly ('true volume') and with different imaging techniques. Three-dimensional echocardiography (3D-Echo) and magnetic resonance imaging (MRI), both of which permit a volume estimation without necessitating geometric assumptions, and angiocardiographic volumetry which is based on the Simpson rule were used in this study.
Results: The best results were achieved with MRI (RV: r(2)=0.99, mean difference: -1. 9+/-3.3%; LV: difference r(2)=0.99,: 2.9+/-5.0%). Likewise, 3D-Echo showed a very good correlation with the true volumes (RV: r(2)=0.93, difference: 9.3+/-6.3%; LV r(2)=0.96, difference: 4.8+/-9.9%). The greatest deviations were observed during angiocardiographic volumetry (LV: r(2)=0.98; difference: 14.4+/-9.2%), particularly when measuring the right ventricle (RV: r(2)=0.82, difference: 57. 9+/-40.1%). Consequently, the direct comparison between 3D-Echo and the other methods yielded the best correspondence with MRI (RV: Bias: 3.7 ml, limits of agreement: 7.7 ml; LV: Bias: 3.7 ml, limits of agreement: 4.9 ml). In contrast, the differences between 3D-Echo and angiocardiography were marked (RV: Bias: 25.5 ml, limits of agreement: 11.1 ml; LV: Bias: 8.7 ml, limits of agreement: 13.2 ml).
Conclusion: In a porcine cardiac model, 3D-Echo permits a relatively precise measurement of ventricular volumes with a slight under-estimation. MRI yielded the most precise volumetry, and the correlation between 3D-Echo and MRI was quite good. Particularly for the right ventricle, the angiocardiographic measurement was attached with the greatest error and thus appears ill-suited for the volumetry of geometrically more complex ventricles.