Based On Single Pilot Study On The Method Of Ecg Signal Detection And Extraction Of Atrial Fibrillation

Atrial fibrillation (AF) is a human heart disease common in arrhythmias. The study of AF electrophysiological particularly the detection of AF and the extraction of AF signal from ECG is of great significance in clinical medical research.This paper presented a new AF detect method for one-lead ECG signal. A modified first-derivative based QRST detection algorithm as well as a proposed adaptive f wave number detection algorithm was utilized in the method, and new conditions was set for AF detection. Different kinds of actual ECG data was used in the experiment to validate this method, including one-lead ECGs of normal, AF and continuous ventricular arrhythmia which collected by local hospital and MIT database. The experimental results showed the accuracy, sensitivity and positive predictive value were96.25%,96.25%and100%, which means the new method had a better performance:distinguishing the continuous ventricular arrhythmia and AF successfully, while achieving high accuracy in AF detection.Meanwhile, this paper presented a new indirect algorithm for extraction of AF signal from one-lead ECG signal. There were two stages in the algorithm. The mean square cross prediction error (MSCPE) based blind source extraction algorithm was utilized to extract the main components of ECG in the first stage. Then, a modified spectral subtraction approach was proposed to extract the atrial fibrillation signal. The one-lead AF ECGs from patients with persistent AF collected by Local hospital was used in the experiment to validate this method. Experimental results showed that the averaged spectral concentration (SC) of extracted atrial fibrillation signal was66.04%while its averaged main frequency peak (fp) was4.0313Hz, and the proposed algorithm had a better performance compared with other algorithms.