What ICA provides for ECG signal extraction from contaminated ECG observations without using differential amplifiers

• Published in: 2010 International Conference on Information and Emerging Technologies pp. 1 - 5
• Main Authors: Ullah, K, Khan, Mohammad A U, Ullah Kundi, Rehmat
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2010
• Subjects: Differential amplifiers; ECG; Electrocardiography; Electrodes; Heart; Independent Component Analysis; JADE algorithm; Kurtosis; Leg; Noise; Software
• ISBN: 9781424480012; 1424480019
• DOI: 10.1109/ICIET.2010.5625697

ECG signals are used very widely as a clinical tool to determine the rate and regularity of heartbeats as well as the size and position of the chambers, the presence of any damage to the heart, and the effects of drugs or devices used to regulate the heart. For accurate assessment of heart the recorded ECG signal should be free of noises and artifacts. Routinely recorded ECG, using hardware differential amplifiers are often corrupted by a drift due to imperfect virtual ground and also it can't totally remove the common mode noise due to its Common Mode Rejection Ratio(CMRR) limitations. This paper introduces a software alternative of an essential hardware differential amplifier component in term of Independent Component Analysis (ICA) to remove the common mode and other noises. The observations from left arm and right arm of the subjects with respect to right leg are acquired using leads, and recorded data was then entered to the model, to estimate demixing matrix of the ICA algorithm. Once this demixing matrix is obtained it was put into the model and thus ECG and noise was separated as independent components. Our model extracts the ECG signal from contaminated signal with variance 0.002 and the kurtosis 12.34 while for filtering method these are 0.23 and 16.75 respectively. With differential filtering technique the SNR is 23dB, in our case it is 27.65. The results of this new model were compared with other existing models and it was found that our model has smaller variance, kurtosis and greater SNR ratio. This model may be helpful in the clinical diagnosis of heart patients.