ECG Signal Denoising by Using Least-Mean-Square and Normalised-Least-Mean-Square Algorithm Based Adaptive Filter Uzzal Biswas*, Anup Das, Saurov Debnath, and Isabela Oishee Electronics and Communication Engineering Discipline Khulna University, Khulna – 9208, Bangladesh *ujjal_ku_07@yahoo.com Abstract— Electrocardiogram (ECG) is a method of measuring the electrical activities of heart. Every portion of ECG is very essential for the diagnosis of different cardiac problems. But the amplitude and duration of ECG signal is usually corrupted by different noises. In this paper we have done a broader study for denoising every types of noise involved with real ECG signal. Two adaptive filters, such as, least-mean-square (LMS) and normalized-least-mean-square (NLMS) are applied to remove the noises. For better clarification simulation results are compared in terms of different performance parameters such as, power spectral density (PSD), spectrogram, frequency spectrum and convergence. SNR, %PRD and MSE performance parameter are also estimated. Signal Processing Toolbox built in MATLAB ® is used for simulation, and, the simulation result clarifies that adaptive NLMS filter is an excellent method for denoising the ECG signal. Keywords- Noises; ECG signal; Adaptive filter; SNR; %PRD; PSD; Spectrogram. I. INTRODUCTION ECG is generated by the heart muscle and measured on the skin surface of the body. When the electrical abnormalities of the heart occur, the heart cannot pump and supply enough blood to the body and brain. As ECG is a graphical recording of electrical impulses generated by heart, it is needed to be done when chest pain occurred such as heart attack, shortness of breath, faster heartbeats, high blood pressure, high cholesterol and to check the heart’s electrical activity. An ECG is very sensitive, different types of noise and interference can corrupt the ECG signal as the real amplitude and duration of the signal can be changed. ECG signals are mostly affected by white noise, colored noise, electrode movement noise, muscle artifact noise, baseline wander, composite noise and power line interference. These noise and interference makes the incorrect diagnosis of the ECG signal [1-3]. So, the removal of these noise and interference from the ECG signal has become very crucial. Different types of digital filters (FIR and IIR) have been used to solve the problem [3-5]. However, it is difficult to apply these filters with fixed coefficients to reduce different types of noises, because the ECG signal is known as a non-stationary signal. Recently, adaptive filtering has become effective and popular methods for processing and analysis of the ECG signal [6-8]. It is well known that adaptive filters with least mean square (LMS) algorithm show good performance for processing and analysis of signal which are non stationary [1]. And in this study, we have used adaptive LMS and normalized least mean square (NLMS) filter to denoise the ECG signal. We also have evaluated their performance. But it is shown that NLMS filter removes all specified noise (mentioned above) more significantly. II. MATERIALS AND METHODS The original ECG signal is taken from the MIT-BIH arrhythmia database [9]. The different types of noise signal are generated by using MATLAB ® . The noise signal is then added with the real ECG signal. To remove the different types of noises, the noisy ECG signal is then pass through two adaptive filter algorithms (e.g., LMS and NLMS). However, the basic block diagram for understanding the overall adaptive filtering process is depicted in Fig. 1. Figure 1. Principle of adaptive filter[7]. The block diagram indicates that, if the value of N(n) is known, then after subtracting this from the mixed signal d(n), the original signal X(n) is obtained. But it is difficult due to the harmonics of noise signal. For this reason an estimated noise signal N´(n) is calculated through some filters and measureable noise source S(n). If N´(n) is more close to N(n), then the estimated desired signal is X´(n) more close to the original signal X(n). Mathematically the output is given by eൌX൅Nെy ሺͳሻ Digital Filter y(n) = N′(n) Adaptive Algorithm ∑ - S(n) d(n) = X(n) + N(n) + e(n) = X′(n) 3rd INTERNATIONAL CONFERENCE ON INFORMATICS, ELECTRONICS & VISION 2014 978-1-4799-5180-2/14/$31.00 ©2014 IEEE