Separating heart sound from lung sound using Non-Local Means filter

Primary aim of this work is to develop methodology for extracting useful information from the phonocardiographic signal for a computerized cardiac auscultation system and the intelligent stethoscope. The paper presents a technique based on a Non-Local Means method (NLM). When synthetic data is combined with the heart sound signal, it is found that NLM can separate the heart sound without degrading the main characteristic features of heart sound at wide diapason of signal-to-noise ratio (SNRin). The results of application of NLM to real auscultation signal are also promising.