Assessment of right ventricular function by 16-detector-row CT: comparison with magnetic resonance imaging.

Abstract

The purpose of this study was to determine right ventricular (RV) function from 16-detector-row CT by using two different software tools in comparison with MRI. Nineteen patients underwent cardiac CT. (1) With semiautomated contour detection software end-diastolic and end-systolic RV volumes were determined from short-axis CT reformations (MPR) created at every 10% of the RR-interval. (2) End-systolic and end-diastolic axial images were transformed to 3D to determine the volumes by using a threshold-supported reconstruction algorithm. Steady-state free-precession cine-MRI of the heart was done in short-axis orientation. RV function could not be analyzed in one patient because of sternal wire artifacts in MRI. Mean end-diastolic (155.4+/-54.6 ml) and end-systolic (79.1+/-37.0 ml) RV volumes determined with MPR correlated well with MRI [151.9+/-53.7 ml (r=0.98) and 75.0+/-36.0 ml (r=0.96), respectively (P<0.001)]. RV stroke volume (76.2+/-20.2 ml for MPR-CT, 76.9+/-20.7 ml for MRI, r=0.93) showed a good correlation and RV ejection fraction (50.8+/-8.4% for MPR-CT, 51.9+/-7.4% for MRI, r=0.74) only a moderate one. Threshold supported 3D reconstructions revealed insufficient correlations with MRI (r=0.31-0.59). MPR-based semiautomated analysis of cardiac 16 detector-row CT allows for RV functional analysis. The results correlate well with MRI findings. Threshold value-supported 3D reconstructions did not show satisfying results because of inhomogeneities of RV contrast enhancement.