Adaptive-Filtering-Based Algorithm for Impulsive Noise Cancellation from ECG Signal

Suppression of noise and artifacts is a necessary step in biomedical data processing. Adaptive filtering is known as useful method to overcome this problem. Among various contaminants, there are some situations such as electrical activities of muscles contribute to impulsive noise. This paper deals with modeling real-life muscle noise with α-stable probability distribution and adaptive filtering noise cancellation assessment with maximum correntropy criterion (MCC) as adaptive technique. Based on our test on some data of MIT-BIH arrhythmia and EMBC databases, we achieve an improved SNR in any electrocardiogram (ECG) signal corrupted by impulsive noise. The worst achieved improvement based on setting the best parameter values using trial and error for both filter and utilized algorithm is 9.5 dB with correlation coefficient value of 0.93. The SNR improvement on the whole utilized database records is 11.03 dB on average. The proposed algorithm is applied to the records from MIT-BIH arrhythmia and EMBC databases to remove the impulsive noise. A computer simulation is used to create and add it to the ECG signals. Simulation results are also provided to support the discussions.