Comparative analysis of Lung sound denoising technique

Pulmonary disorders are causing a huge increments in the mortality rate. The main reason behind this is misinterpretation or delay in diagnosis of disease. Developing a computerized system which can help clinicians to remove this problem will give a better result in controlling these adverse effects of increasing mortality rate. Data collection and their preprocessing with the help of computerized system will tend to remove the chances of overlapping of results which may cause misinterpretation. For developing a computerized system. Data from 15 patients have been collected in the time period of one month. Along with the results of pulmonary function test the overlapping are easily visible which can be overcome by including heart sound and lung sound in our analysis. However, lung sound signals are subjected to different kind of noises. Therefore, six denoising techniques are Wavelet, Savitzky Golay Moving average filter, FIR, Median filter and Butterworth filter are implemented and evaluated. The performance of all the three filters are compared on the basis of signal to noise ratio. Wavelet denoising technique gives better result with a Signal to noise (SNR) value of 84.43dB.