Digital Signal Processing Algorithms and Techniques for the Enhancement of Lung Sound Measurements

Lung sound signal (LSS) measurements are taken to aid in the diagnosis of various diseases. Their interpretation is difficult however due to the presence of interference generated by the heart. Novel digital signal processing techniques are therefore proposed to automate the removal of the heart sound signal (HSS) interference from the LSS measurements. The HSS is first assumed to be a periodic component so that an adaptive line enhancer can be exploited for the mitigation of the HSS interference. The utility of the scheme is verified on synthetic signals, however its performance is found to be limited on real measurements due to sensitivity in the selection of a decorrelation parameter. An improved solution with multiple measurements, that does not require a decorrelation parameter and exploits the spatial dimensions, is therefore proposed on the basis of blind source extraction based upon second-order statistics. This approach is found to have improved performance on both real and synthetic datasets, although the level of departure from true periodicity impacts this improvement. A new sequential blind extraction algorithm for removing quasi-periodic signals with time-varying period is then developed. Source extraction is performed by sequentially converging to a solution which effectively diagonalizes autocorrelation matrices at time lags corresponding to the ii Abstract iii time-varying period, and thereby exploits a key statistic of the nonstationary desired interfering source. The algorithm is shown to have fast convergence and to yield much improvement in signal-to-interference ratio (SIR) as compared to when a fixed period is assumed. Separation of the HSS interference is confirmed on measurement datasets. To conclude, a complete algorithmic solution for the removal of the HSS interference from the LSS measurements, incorporating automatic peak detection based on particle-filtering to extract the time-varying period of the HSS interference, is proposed and validated on real-world lung sound recordings.iii time-varying period, and thereby exploits a key statistic of the nonstationary desired interfering source. The algorithm is shown to have fast convergence and to yield much improvement in signal-to-interference ratio (SIR) as compared to when a fixed period is assumed. Separation of the HSS interference is confirmed on measurement datasets. To conclude, a complete algorithmic solution for the removal of the HSS interference from the LSS measurements, incorporating automatic peak detection based on particle-filtering to extract the time-varying period of the HSS interference, is proposed and validated on real-world lung sound recordings. To my loving...... and to my parents