Abstract 2236: Automatic Endocardium Recognition and Tracking System For Automated Quantitation of Left Ventricular Ejection Fraction: Clinical Reproducibility and Accuracy of AutoEF
Background and Purpose: Quantitative approaches to LV EF have been unpopular because among other reasons they entail manual tracing of endocardial borders. We evaluated whether AutoEF (Siemens, CA) a fully automated endocardial recognition and tracking algorithm was accurate and efficient in measurement of EF.
Methods: 94 patients underwent 2-D echocardiogram examinations: 44 with EF > 50% and 50 with LVEF < 50% - EF measured with Bi-Plane Simpson’s (manual) by an expert echocardiographer. LVEF was also measured with an Auto-EF package where the operator merely identified the 4C and 2C views. MRI was performed on 51 patients: 22 with EF< 50 by manual echo method. The time taken to measure LVEF by manual and AutoEF was recorded.
Results: LVEF by Auto-EF strongly correlated with manual estimate (n = 94, r = 0.82, p < 0.01) and MRI measurements (n = 31, r = 0.87, p < 0.001) of EF. Accuracy of AutoEF compared to MRI was excellent irrespective of EF: r = 0.65, p = 0.02 for EF > 50% and r = 0.71, p = 0.007 for EF > 50 %. Variability of EF measures by AutoEF and manual methods were: 6 ± 2 % (p = NS) and 11 ± 3 % (p = 0.01) respectively. The time taken by AutoEF and manual method was 33 ± 9 sec and 226 ± 54 sec, respectively (14% difference, p < 0.05).
Conclusions: Auto-EF is an accurate, efficient and reproducible 2-D Echo method to quantify LVEF. This may afford an excellent incentive to measure rather than estimate EF.