Reliability and reproducibility of quantitative assessment of left ventricular function and volumes with 3-slice segmentation of cine steady-state free precession short axis images

Abstract

ObjectivesQuantitative assessment of left ventricular (LV) functional parameters in cardiac MR requires time-consuming contour tracing across multiple short axis images. This study assesses global LV functional parameters using 3-slice segmentation on steady state free precision (SSFP) cine short axis images and compares the results with conventional multi-slice segmentation of LV. MethodsData were collected from 61 patients who underwent cardiac MRI for various clinical indications. Semi-automated cardiac MR software was used to trace LV contours both at multiple slices from base to apex as well as just 3 slices (base, mid, and apical) by two readers. Left ventricular ejection fraction (LVEF), LV volumes, and LV mass were calculated using both methods. ResultsBland–Altman plot revealed narrow limits of agreement (−4.4% to 5.1%) between LVEF obtained by the two methods. Bland–Altman analysis showed slightly wider limits of agreement between end-diastolic volumes (−5.0 to 12.0%; −3.9 to 8.5ml/m2), end-systolic volumes (−10.9 to 14.7%; −4.1 to 6.5ml/m2), and LV mass (−5.2 to 12.7%; −4.8 to 10.2g/m2) obtained by the two methods. There was a small mean difference between LV volumes and LV mass obtained using multi-slice and 3-slice segmentation. No statistically significant difference existed between the LV parameters obtained by the two readers using 3-slice segmentation (p>0.05). Multi-slice assessment required approximately 15min per study while 3-slice assessment required less than 5min. Conclusions3-slice segmentation of the left ventricle at basal, mid, and apical levels on cine SSFP short axis images can provide rapid and reliable assessment of LVEF with good reproducibility. The 3-slice method also provides a reasonable estimate of the LV volumes and LV mass.