RESUMO
Heart sound signals acquisition is the primary basis for achieving non-invasive diagnosis of coronary heart disease. In this paper, a digital signal processor (DSP)-based on miniaturized circuit of heart sound signals acquisition and analysis platform was designed to achieve the functions of filtering, collecting, processing, displaying and the communicating with PC. With the self-developed experimental platform, we collected 228 cases of heart sounds of clinical data, and processed the signals using de-noising method with wavelet transform. These experimental results indicated that the db6 wavelet has the most obvious de-noising effect among the four most commonly used wavelets, i.e., haar, db6, sym8, and coif5. One wavelet at different levels possessed different de-noising effects, with level-5 db6 decomposition obtaining the most desirable result.