Objectives: Visual assessment of the ejection fraction (EF) is often used in clinical practice, but is subjective and requires training and experience. The auto EF method has been newly developed for automated detection of the left ventricular (LV) endocardial border throughout the cardiac cycle. The clinical feasibility was assessed of the auto EF method for measuring LV volumes and EF in patients. Manually traced EF measured by the biplane modified Simpson's rule was used as the reference standard.
Methods: The study population consisted of 30 consecutive patients with normal sinus rhythm. All patients underwent two-dimensional echocardiography. The auto EF method incorporated pattern and shape recognition to automatically locate the LV, track the endocardium, and calculate EF from routine digital images. LV end-diastolic and end-systolic volumes and EF measured by the auto EF method were compared those by the manually traced method. LVEF by the Auto EF method was also compared with visual EF determined by nonattending physicians.
Results: Auto EF was more reproducible than visual EF by the expert reader. LVEF by the auto EF method had excellent correlation and close limits of agreement with manually traced EF compared with visual EF(4-chamber view: y = 1.10 x - 4.28, r = 0.94, bias = 1.6%, 2-chamber view: y = 0.88 x + 7.25, r = 0.90, bias = 0.4%). LV volumes by the auto EF method underestimated those by manually traced EF (end-diastolic volume by 4-chamber view: y = 0.72 x + 8.92, r = 0.93, bias = - 16.7 ml, end-systolic volume by 4-chamber view: y = 0.82 x - 0.97, r = 0.98, bias = - 8.5 ml, end-diastolic volume by 2-chamber view: y = 0.73 x + 12.4, r = 0.82, bias = - 12.3 ml, end-systolic volume by 2-chamber view: y = 0.65 x + 6.94, r = 0.90, bias = - 6.8 ml).
Conclusions: The auto EF method is a clinically useful tool for the measurement of LV volumes and EF.