Objectives: Quantification of left ventricular (LV) mass has important prognostic implications. However, accurate measurement of LV mass has been difficult, in part because of the oblique angle at which the heart lies within the chest and the continuous movement of the heart itself. Multislice computed tomography (MSCT) allows assessment not only of coronary stenosis but LV volume, function, and mass. A novel three-dimensional (3D) region-growing-based semi-automated segmentation algorithm for measurements of LV mass, volume, and function was recently developed. This study evaluated this new 3D automated method for measurement of LV mass, by comparison with a well-established 2D manual contour-drawing algorithm.
Methods and results: The study population consisted of 50 consecutive patients who underwent ECG-gated MSCT for evaluation of coronary arteries. The 3D algorithm for reliable segmentation was unsuccessful in two patients. In the remaining 48 patients, however, LV segmentation using this algorithm was performed and delivered visually reliable segmentation results. The 3D algorithm for analysis of LV function and mass is feasible based on volumetric data, and exhibits good correlation and agreement with the results obtained with the conventional 2D algorithm. The time required for the new automated algorithm was significantly shorter than that for the manual contour-drawing algorithm (P<0.0001) (automated algorithm: 468.0±205.1 s, manual algorithm: 1362.4±410.5 s, mean ±S.D.).
Conclusions: The 3D semi-automated region-growing segmentation algorithm for analysis of LV function and mass is feasible based on volumetric data, and exhibits good correlations and agreement with the results of the conventional 2D manual contour-drawing algorithm.