AIMS: The feasibility and accuracy of an automated adaptive algorithm (Heart Model) for the quantification of left heart chamber volumes and left ventricular ejection fraction has been reported earlier. An improved version of the algorithm is available, and we aimed to test its accuracy compared with manual 3D echocardiography. METHODS AND RESULTS: Apical 3D transthoracic datasets were obtained in 67 patients. Acquisitions covering ventricles and atria were performed for analysis using the automated software (Heart Model, Philips Healthcare). In addition, acquisitions focused on the left ventricle were acquired for left ventricle manual 3D measurements (QLAB 10 3DQA, Philips Healthcare). Automated results using endocardial contouring settings at 50% showed a strong correlation with manual 3D measurements (r=.84-.97). Left ventricular end-diastolic volumes were underestimated (bias -5.9 mL, LOA ±38.5 mL), with no significant differences in other parameters. Intra-observer variability using the automated algorithm was zero for all parameters given the lack of human interaction. Manual corrections of the automatic algorithm introduced small but significant differences in volumes but not in ejection fraction when compared with automatic results. Manually corrected results of the automatic algorithm were not significantly different from those obtained with manual 3D echocardiography, except for a small underestimation of left atrial volumes (bias -2.1 mL, LOA ±15.2 mL). CONCLUSIONS: The current improved version of the automated adaptive algorithm is accurate for the assessment of left heart chamber volumes, albeit a small underestimation of left ventricular end-diastolic volume is seen, when compared with manual 3D echocardiography.
AIMS: The feasibility and accuracy of an automated adaptive algorithm (Heart Model) for the quantification of left heart chamber volumes and left ventricular ejection fraction has been reported earlier. An improved version of the algorithm is available, and we aimed to test its accuracy compared with manual 3D echocardiography. METHODS AND RESULTS: Apical 3D transthoracic datasets were obtained in 67 patients. Acquisitions covering ventricles and atria were performed for analysis using the automated software (Heart Model, Philips Healthcare). In addition, acquisitions focused on the left ventricle were acquired for left ventricle manual 3D measurements (QLAB 10 3DQA, Philips Healthcare). Automated results using endocardial contouring settings at 50% showed a strong correlation with manual 3D measurements (r=.84-.97). Left ventricular end-diastolic volumes were underestimated (bias -5.9 mL, LOA ±38.5 mL), with no significant differences in other parameters. Intra-observer variability using the automated algorithm was zero for all parameters given the lack of human interaction. Manual corrections of the automatic algorithm introduced small but significant differences in volumes but not in ejection fraction when compared with automatic results. Manually corrected results of the automatic algorithm were not significantly different from those obtained with manual 3D echocardiography, except for a small underestimation of left atrial volumes (bias -2.1 mL, LOA ±15.2 mL). CONCLUSIONS: The current improved version of the automated adaptive algorithm is accurate for the assessment of left heart chamber volumes, albeit a small underestimation of left ventricular end-diastolic volume is seen, when compared with manual 3D echocardiography.
Authors: David G Platts; Manan Vaishnav; Darryl J Burstow; Christian Hamilton Craig; Jonathan Chan; John L Sedgwick; Gregory M Scalia Journal: Int J Cardiol Heart Vasc Date: 2017-09-03