A Study Of Lung Sound Extraction Based On Blind Source Separation

Lung sound signal is a biological signal, which is the body's respiratory system with the outside world resulting in the ventilation process. Lung sound signal forming complex but contains a wealth of physiological and pathological information. In modern medicine, electronic auscultation is still an important and common method to diagnose respiratory diseases. However, in the process of auscultation, heart sounds is an inevitable interference. Removing heart sounds from lung sound recordings is a challenging and hot task but of great interest. In this paper, we using blind source separation (BSS) technique to separate heart sound signal from lung sound signal.The so-called blind source separation is that mixing process and being mixed source signals are in unknown circumstances, according to a small amount of prior information to recover and estimate the source signals. The prior information is often the basic assumptions(Source signals are statistically independent) of the blind signal separation. The problem of blind source separation that linear real-time mixing and under the assumption of independence is also known as independent component analysis (ICA). The statistical independence of source signals is a very loose condition, so blind signal separation has been widely used in many areas such as signal processing, image signal processing, biomedical signal processing, communications, radars and so on.This paper introduces basic mathematical principles and the basic model, the steps for solving problems, theory and algorithm of the independent component analysis are elaborated and analysised, the application of ICA in signal processing also discussed, remain problems are proposed and the direction of future research are indicated. The related issues of implementation of speech separation with the algorithms of fast independent component analysis (FastICA algorithm) and the maximum amount of information algorithm (Infomax algorithm) are detailed. Finally, the fact that heart sound and lung sound signal is a short-time stationary non-stationary signal we use Infomax algorithm and FastICA algorithm to achieve separation of and lung sound signals and heart sound signals, and finally througth emulational experiments, they are proved to be of better performance, compared with other algorithms. Finally, the simulation experiments show that these two algorithms have better performance. With the lung sound signal with the previous extraction methods, we use blind source separation method, can be isolated from lung sound signal, but also isolated the heart sound signal. Therefore, this research has a dual meaning.