Computer based analysis for heart and lung signals separation

In this paper, two methodologies are proposed to enhance the automatic noise cancellation and signal separation between heart and lung sounds. In fact, transient signals such as heart and lung signals may undergo abrupt or sharp change in the first and second derivatives. A real separation between such two interfering mixed signals needs an efficient approach to avoid losing important information in both signals. Rhythmic cardiac signal contains important characteristics which can be exploited to develop adaptive based algorithms that allow efficient separation between lung and heart signals when they are mixed in a recorded signal. In the first proposed methodology we have developed an algorithm based on adaptive filtering technique and build using multiple filtering functions with coefficients correlated to the mixed source signal. In the second methodology, fast independent component analysis was developed to cancel heart sound in lung mixed sound. Both methods are well detailed in this work, and a comparative study is achieved to evaluate the efficiency of each method. A high accuracy of the new proposed algorithms is found and many applications are used to quantify the performances of these techniques.