Semi-automatic measurement of left ventricular function on dual source computed tomography using five different software tools in comparison with magnetic resonance imaging

Abstract

To compare left ventricular (LV) function assessment using five different software tools on the same dual source computed tomography (DSCT) datasets with the results of MRI. Twenty-six patients, undergoing cardiac contrast-enhanced DSCT were included (20 men, mean age 59±12 years). Reconstructions were made at every 10% of the RR-interval. Function analysis was performed with five different, commercially available workstations. In all software tools, semi-automatic LV function measurements were performed, with manual corrections if necessary. Within 0-22 days, all 26 patients were scanned on a 1.5 T MRI-system. Bland-Altman analysis was performed to calculate limits of agreement between DSCT and MRI. Pearson's correlation coefficient was calculated to assess the correlation between the different DSCT software tools and MRI. Repeated measurements were performed to determine intraobserver and interobserver variability. For all five DSCT workstations, mean LV functional parameters correlated well with measurements on MRI. Bland-Altman analysis of the comparison of DSCT and MRI showed acceptable limits of agreement. Best correlation and limits of agreement were obtained by DSCT software tools with software algorithms comparable to MRI software. The five different DSCT software tools we examined have interchangeable results of LV functional parameters compared to regularly analysed results by MRI. The best correlation and the narrowest limits of agreement were found when the same software algorithm was used for both DSCT and MRI examinations, therefore our advice for clinical practice is to always evaluate images with the same type of post-processing tools in follow-up.