Simulations To Remove Respiratory Signals From ECG By Using Independent Component Analysis

High-quality electrocardiogram(ECG) is very important and is widely practical in the clinical practice. But ECG signals are so sensitive to many factors such as the respiratory signal. De-noising the respiratory signal from ECG can highly improve the quality of ECG. Now, many methods have been used in this field. Here, a new method named as the Independent Component Analysis (ICA)is proposal in the project.ICA is a signal processing technique for multi-dimensional signals to be separated into the independent original signals. That comes from the Blind Source Separation(BSS). The paper investigates to remove respiratory signals from ECG by using ICA algorithm.The work is done in the fallowing aspects:1. Four ICA algorithms are used to de-noise the signals in the case that the separated signals are produced by the simulated ECG signals mixing with an ideal noisy signal in linear manner. Here, the simulated ECG signals are formed by three dynamical equations. Both of the convergence rate and the correlated parameter are used to evaluate the ICA algorithms.2. Four ICA algorithms are used to separate the respiratory signal from the signals that are produced by the simulated ECG signals mixing with a real respiratory signal sampled in PowerLab Signal Sample System in non-linear manner. And, both of the convergence rate and the correlated parameter are used to evaluate the ICA algorithm to choose a better ICA algorithm in the later work.3. The ICA algorithm chosen in step2 is used to remove the respiratory signal from the real signals that the ECG signals contaminated by the respiratory signal and the respiratory are all sampled in PowerLab Signal Sample System at same time. The correlated parameter is used to evaluate the effect of the separation.It is concluded that the ICA algorithms is effective and advance in de-noising the respiratory signal, not needing much more pre-acknowledge.