Title :
Quasi-periodic atrial activity components in the ECG used to discriminate between paroxysmal and chronic atrial fibrillation
Author :
Lemay, M. ; Dang, L. ; Vesin, JM
Author_Institution :
Inst. of Electr. Eng., Ecole Polytech. Fed. de Lausanne, Lausanne
Abstract :
Spatiotemporal organization in atrial fibrillation has recently been observed in invasive studies with a left-to-right frequency gradient. We propose the use of a recently developed technique named phase-rectified signal averaging to estimate the mean activation rates in leads V1 and V5 to observe the same left-to-right gradient in a noninvasive manner. Based on these values, a classification procedure between paroxysmal (n=43) and chronic (n=20) atrial fibrillation patient is suggested. The processing steps were: baseline correction and ventricular activity cancellation, followed by the phase-rectified signal averaging technique. The three features used were the dominant frequency values of leads V1 and V5 and the absolute value of the difference between these frequencies. These yielded to 84.1% of correct classifications (p < 8.5 times 10-7).
Keywords :
bioelectric phenomena; cardiovascular system; diseases; electrocardiography; medical signal processing; signal classification; spatiotemporal phenomena; ECG; baseline correction; chronic atrial fibrillation; frequency gradient; paroxysmal atrial fibrillation; phase-rectified signal averaging technique; quasiperiodic atrial activity component; signal classification; spatiotemporal organization; ventricular activity cancellation; Atrial fibrillation; Databases; Electrocardiography; Frequency; Humans; Inspection; Phase estimation; Protocols; Spatiotemporal phenomena; Yield estimation;
Conference_Titel :
Computers in Cardiology, 2008
Conference_Location :
Bologna
Print_ISBN :
978-1-4244-3706-1
DOI :
10.1109/CIC.2008.4749168