Currently, the left ventricular ejection fraction (LVEF) is determined by invasive left ventriculography during angiography . However, this method can have risks, including contrast exposure. To avoid such complications and automate the prediction of LVEF from angiograms, researchers tested a video-based deep neural network (DNN) called CathEF. About 4,042 adult angiograms with corresponding transthoracic echocardiograms (TTE) measured LVEF from 3,679 patients were used in the study.
CathEF effectively discriminated reduced LVEF (£40%) and predicted continuous LVEF from angiogram videos of the left coronary artery. However, CathEF was found to underestimate high LVEFs and overestimate low LVEFs. Early estimation of LVEF from angiogram videos of the left coronary artery by using a DNN can be a potential non-invasive procedure to measure heart function. However, extreme measures of the…
