Partial voxel interpolation to reduce partial volume error of cardiac computed tomography ventricular volumetry in patients with congenital heart disease

Abstract

Background Varying degrees of partial volume error depending on the complexity of the endocardial borders are inevitable in threshold-based cardiac computed tomography (CT) ventricular volumetry. These errors can potentially be reduced by using a partial voxel interpolation (PVI) method, but this has not been tested for cardiac CT ventricular volumetry.

Objective To evaluate the partial volume error-reducing effects of the PVI method in cardiac CT ventricular volumetry among patients with congenital heart disease (CHD).

Materials and methods The cardiac CT ventricular volumetry data were obtained from 55 patients (median age 12.0 years) with CHD. The ventricular and myocardial volumes, ejection fraction and ventricular mass-volume ratio were quantified and compared before and after the PVI method. The correlation between the myocardial volumes in the end-systolic and end-diastolic phases was tested. The effect of the PVI method on the classification of ventricular hypertrophy was evaluated.

Results The indexed ventricular volumes after PVI were significantly smaller (7.4–11.5%) than those before PVI (P<0.001). In contrast, the indexed myocardial volumes were significantly larger (6.2–27.7%) after PVI (P<0.001). The ejection fractions and mass-volume ratios were significantly larger (1.6–2.2% and 19.7–42.5%, respectively) after PVI (P<0.001 and P<0.001, respectively). The indexed myocardial masses showed prominently high correlation between the end-systolic and end-diastolic phases (R, 0.961–0.990; P<0.001). The proportions of no and severe hypertrophy were significantly decreased (P<0.002) and increased (P<0.032), respectively, after the application of the PVI method.

Conclusion The PVI method can reduce partial volume error in cardiac CT ventricular volumetry among patients with CHD.