Evaluation of new software for angiographic determination of right ventricular volumes.

OBJECTIVES: The scope of the study was to evaluate a recent software for angiographic volume determination as compared to cardiovascular magnetic resonance imaging.
BACKGROUND: A new right ventricular analysis software closes a diagnostic gap in quantitative angiography. Cardiovascular magnetic resonance imaging short axis multi slice summation is a validated reference standard.
METHODS: Right ventricular angiograms were acquired in frontal and lateral projection in 15 pediatric and 17 adult patients. Additional angiograms were acquired in RAO30 degrees /LAO60 degrees projections in 10 adult patients. The tested models comprised area length with different regressions, multi-slice with different regressions, Boak, and pyramid method. Original regressions were used to calculate angiographic right ventricular volume. Right ventricular reference volumes were determined by multi-slice summation from cardiac magnetic resonance short-axis images.
RESULTS: Mean inter-observer difference was -1 ml (95% confidence: -35-34 ml) and mean intra-observer difference was 0 ml (95% confidence: -22-22 ml). There was no significant difference (4 ml, 95% confidence: -22-30 ml) between geometric calibration and calibration by a sphere. The area length and multi-slice models demonstrated the best agreement with the cardiac magnetic resonance reference. Performance was best for the Onnasch Lange models.
CONCLUSION: The evaluated software provides acceptably accurate volume estimates for the majority of ventricles. In a few cases larger errors may occur, however. The area length and multi-slice models preferably with Onnasch Lange regressions may be recommended. Inter- and intra-observer agreement were excellent. Geometric calibration using data from DICOM header files may be used.