International Journal of Electrical and Computer Engineering (IJECE) Vol. 3, No. 3, June 2013, pp. 307~316 ISSN: 2088-8708 ? 307 Journal homepage: http://iaesjournal.com/online/index.php/IJECE Analysis of Noise Sensitivity of Different ECG Detection Algorithms A. B. M. Aowlad Hossain*, M. A. Haque** *Department of Electronics and Communication Engineering, Khulna University of Engineering & Technology **Department of Electrical and Electronic Engineering, Bangladesh University of Engineering & Technology Article Info ABSTRACT Article history: Received Feb 27, 2013 Revised Apr 18, 2013 Accepted May 18, 2013 This paper presents an analysis of noise sensitivities of different detection algorithms for electrocardiogram (ECG) taken from MIT-BIH arrhythmia database. Seven methods used in this paper are based on derivatives, digital filters (DF), neural network (NN) and wavelet transform (WT). The raw ECG is corrupted with 5 different types of synthesized noise, namely, power line interference, base line drift due to respiration, abrupt baseline shift, electromyogram (EMG) interference and a composite noise made from other types. A total of 315 data sets are constructed from 15 raw data sets for each type of noise adding 0%, 25%, 50%, 75% and 100% noise levels. The application of the methods to detect QRS complexes of a total of 33,774 beats of ECG shows that none of the algorithms are able to detect all QRS complexes without any false positives for all of the noise types at the highest noise level. Algorithms based on NN and WT show better performance considering all noise types and the two algorithms perform similarly. The result of this study will help to develop a more robust ECG detector and this will make ECG interpretation system more effective. Keyword: Electrocardiogram QRS complex detection Noise sensitivity Copyright © 2013 Institute of Advanced Engineering and Science. All rights reserved. Corresponding Author: A. B. M. Aowlad Hossain, Department of Electronics and Communication Engineering, Khulna University of Engineering & Technology, KUET, Khulna-9203, Bangladesh. Email: aowlad0403@yahoo.com 1. INTRODUCTION Electrocardiogram (ECG) is the graphical representation of the bioelectric potential generated by the muscles of the heart. ECG consists of characteristic P wave, QRS complex and T wave. Physicians always look for the amplitudes and duration of the waves as well as the inter-wave duration to take decision about the condition of cardiac and associated physiological systems. Computer based automatic recognition of ECG characteristic points are necessary to help physicians for quick, easy and accurate diagnosis of cardiac conditions. Computerized electrocardiography is now a well-established clinical practice. Because of its specific shape, the QRS complex serves as an entry point for almost all automated ECG detection algorithms. Since QRS portion is usually easy to distinguish due to its relatively high amplitude and sharp peak, simple slope criteria [1], second order derivatives [2], [3] were the major consideration for many ECG beat detection algorithms for the last few decades. The ECG may be masked by many types of noise and interference such as electromyographic (EMG) interference, power line interference, base line drift, abrupt shift in the base line, electrosurgical noise, instrumental noise, electrode contact noise, etc. If an ECG wave is sufficiently corrupted by any of these or other unknown noise types, the signal features may not be readily comprehensible by the visual systems of a human observer. To avoid power line noise and to reduce the influence of the muscle artifact and base line shift, digital filters are used [4]. Filter banks are also used for
