Reduction of Power Line Interference in ECG Using Adaptive Filters: A Comparative Study Between Different Adaptive Noise Cancellation Algorithms

Abstract

The Electrocardiography (ECG) signal carries the information about the electrical activity of the heart, playing very important role in the medical diagnosis of cardiovascular diseases. As the cardiac cells contract, action potentials arise and cause voltage variations, which are measured as ECG signals by surface electrodes placed on the body surface. Unfortunately, large amplitude signals of similar frequency arise and often reach to the skin surface and mix with the ECG signals. So, the ECG may be corrupted by many types of noise such as power line interference (PLI), which can restrict the accuracy of ECG’s heart rate detection algorithms. Several techniques have been proposed over the years to reduce PLI present in the ECG signals like hardware techniques, Notch Filters, Wiener Filters and Adaptive Filters. This paper presents known algorithms of two different approaches to adaptive filters: the NLMS algorithm, based on the classic LMS algorithm and the RLS algorithm. We compared the performance of these algorithms in an adaptive FIR filter structure called adaptive noise cancellation to remove the PLI in ECG signals. The NLMS algorithm needed more time to converge to the lowest squared error, but when it does, this error is lower compared to RLS algorithm. The RLS algorithm, in its turn, has presented a higher computational processing time and a reduced improvement in signal to noise ratio. However, its convergence speed was faster than the NLMS algorithm and it was more efficient in removing superior harmonics from PLI in the noisy ECG signal.