Automatic diagnosis of premature ventricular contraction based on Lyapunov exponents and LVQ neural network. Issue 1 (October 2015)

• Record Type: Journal Article
• Authors: Liu, Xiuling; Du, Haiman; Wang, Guanglei; Zhou, Suiping; Zhang, Hong
• Is Part Of: Computer methods and programs in biomedicine. Volume 122:Issue 1(2015)
• Page Start: 47
• Page End: 55
• Publication Date: 2015-10
• Subjects: PVC -- ECG -- Lyapunov exponents -- LVQ neural network -- Classification; Medicine -- Computer programs -- Periodicals; Biology -- Computer programs -- Periodicals; Computers -- Periodicals; Medicine -- Periodicals; Médecine -- Logiciels -- Périodiques; Biologie -- Logiciels -- Périodiques; Biology -- Computer programs; Medicine -- Computer programs; Periodicals; Electronic journals; 610.28
• DOI: 10.1016/j.cmpb.2015.06.010
• Languages: English
• ISSNs: 0169-2607

Abstract

Abstract : Highlights: Based on the non-uniformity of RR interval, the premature beats can be diagnosed. The features with Lyapunov exponents of ECG beat are extracted. LVQ neural network is used to identify PVC beats. Abstract: Premature ventricular contraction (PVC) is a common type of abnormal heartbeat. Without early diagnosis and proper treatment, PVC may result in serious harms. Diagnosis of PVC is of great importance in goal-directed treatment and preoperation prognosis. This paper proposes a novel diagnostic method for PVC based on Lyapunov exponents of electrocardiogram (ECG) beats. The methodology consists of preprocessing, feature extraction and classification integrated into the system. PVC beats can be classified and differentiated from other types of abnormal heartbeats by analyzing Lyapunov exponents and training a learning vector quantization (LVQ) neural network. Our algorithm can obtain a good diagnostic result with little features by using single lead ECG data. The sensitivity, positive predictability, and the overall accuracy of the automatic diagnosis of PVC is 90.26%, 92.31%, and 98.90%, respectively. The effectiveness of the new method is validated through extensive tests using data from MIT-BIH database. The experimental results show that the proposed method is efficient and robust.