Operator induced variability in cardiovascular MR: left ventricular measurements and their reproducibility

To assess the intra- and inter-operator variability of the manual planning of cardiovascular magnetic resonance imaging and to evaluate the influence of these factors on the functional parameters of the left ventricle (LV).

The study population consisted of 10 healthy volunteers. For each subject the manual planning of the short-axis cine acquisitions was carried out twice by one operator and once by a second operator. Left ventricular volume, mass, and function were manually evaluated twice by one experienced observer, resulting in an approximation of the intra-observer variability factor. The intra- and inter-operator variation factors were estimated as the difference between the total and intra-observer variation components.

LV end-diastolic volume varied by 3.3% and 4.16%, and LV end-systolic volume by 5.84% and 6.23% for intra- and inter-operator studies, respectively. The variability for LV mass at end-diastole was equal to 4.23% in both studies. For the ejection fraction the variability was 3.56% and 2.97% for intra- and inter-operator studies, respectively. Comparison of reproducibility between intra- and inter-operator studies resulted in insignificant statistical differences. Bland-Altman limits of agreements revealed no systematic bias in differences between measurements with respect to their means. Reliability of the planning expressed as the angular deviation of the short-axis imaging planes amounts to 2.67 -/+ 1.5 degrees and 4.99 +/- 2.17 degrees for the intra-operator and inter-operator studies, respectively. For EDV, ESV, and EF approximately 75-80% of the total variation can be explained by the within or between operator variation, while the same percentage is 60% for LVM.

Our study confirms the excellent inter- and intra-operator reproducibility of the cardiovascular magnetic resonance measurements of the left ventricular volumes and mass in a group of healthy volunteers.