Quantification of left ventricular function and mass in cardiac Dual-Source CT (DSCT) exams: comparison of manual and semiautomatic segmentation algorithms.

Abstract

The purpose of our study was to evaluate reliability of left ventricular (LV) function and mass quantification in cardiac DSCT exams comparing manual contour tracing and a region-growing-based semiautomatic segmentation analysis software. Thirty-three consecutive patients who underwent cardiac DSCT exams were included. Axial 1-mm slices were used for the semiautomated technique, and short-axis 8-mm slice thickness multiphase image reconstructions were the basis for manual contour tracing. Left ventricular volumes, ejection fraction and myocardial mass were assessed by both segmentation methods. Length of time needed for both techniques was also recorded. Left ventricular functional parameters derived from semiautomatic contour detection algorithm were not statistically different from manual tracing and showed an excellent correlation (p<0.001). The semiautomatic contour detection algorithm overestimated LV mass (180.30+/-44.74 g) compared with manual contour tracing (156.07+/-46.29 g) (p<0.001). This software allowed a significant reduction of the time needed for global LV assessment (mean 174.16+/-71.53 s, p<0.001). Objective quantification of LV function using the evaluated region-growing-based semiautomatic segmentation analysis software is feasible, accurate, reliable and time-effective. However, further improvements are needed to equal results achieved by manual contour tracing, especially with regard to LV mass quantification.