A study on reproducibility and validation of the measurement of left ventricular ejection fraction

The assessment of left ventricle ejection fraction (LVEF) by gated radionuclide angiocardiography (GRNA) is a widely accepted and validated procedure. However, the problem of inter-observer Variability is still with us. The authors performed a multicenter trial in which each participant was asked to assess the LVEF both manually and automatically on a series of 16 GRNA. Actually, these studies consisted in four groups of GRNA performed in four patients rising four slightly different views : the procedure was blinded to the observers; Manual and automatic processings were compared in terms of reproducibility and robustness. This study showed that, despite ifs operator dependence, manual assessment of IVEF was more reproducible and robust than algorithm determination. On the basis of these results, the authors propose a validation method based on the consensus of skilled operators and raise the problem of the necessity of a formal structure of evaluation.