Noise Removal from ECG Signal Based on Filtering Techniques

In this work, we are focusing on Noise Removal from ECG Signal Based on Filtering Techniques. The Electrocardiogram (EKG or ECG) is a semi-cyclic, rhythmically, and synchronous signal with a cardiac function through the passive sensory apparatus in which the apparatus is performing as generator of bio-electric signal mimicking the function of the heart. The EKG signals are inherently weak, and noisy signals built of many variable components due to many environmental factors in which it may include but is not limited to changes in body temperature, body movement, and line frequency 50/60 Hz. 

The ECG signal cannot be conditioned, amplified, nor reproduced directly and therefore, digital filtering techniques with adjustable window are used in this paper. Our work analyses several models of Finite Impulse Response (FIR) filters of low-pass and high-pass and their aspects in term of response time, gain, and harmonic distortion, and rejection to determine the best band-pass filtering model to reproduce an ECG signal that closely resembles the actual Heart function of a patient. A hybrid filtering model is proposed and experimentally tested. Mean square error (MSE) is used to estimate a signal goodness.

MATLAB environment has been used for the experimental part to simulate the signals. This research work has been considered in the context of a larger project that consists of a complex wearable health monitoring system comprising biosensors, wireless communication modules and links, control and processing units, medical shields, wearable materials and advanced algorithms used for decision making and data extracting. The proposed filtering technique is useful in the medical data preprocessing phase.