A New Automatic Detection Method For Bundle Branch Block Using ECGs

Bundle branch block(BBB)is a type of conduction disturbances of arrhythmia.For patients with BBB,organic heart diseases have significantly high incidence and mortality.Therefore,it is necessary to detect BBB in advance,which is helpful in both control and treatment of related heart diseases.Electrocardiogram(ECG)is one of the most effective ways to diagnose BBB clinically.The traditional approach of ECG-based BBB detection is always achieved by the observation of a trained clinician on whether the corresponding ECG presents the features of BBB.However,the abnormal waveform of ECG with BBB are diversified and weak,which leads to the incomplete and inaccurate results through such an artificial method.Therefore,there has been an increasing interest in the study of the automatic detection method of BBB using ECGs.This paper proposes a novel automatic detection method of BBB.Firstly,a new multiple R-peak detection algorithm is developed;then,two ECG features including the number of R-peak and RR interval are obtained;next,by combining with a linear classification,the automatic detection for BBB is achieved;finally,the performances of the proposed method are verified on the CPSC2018 database,and the results demonstrate that average accuracy,sensitivity and specificity are 96.45%,95.81%,96.80%,respectively.This paper is organized as follows.