Respiratory phase detection based on one lung sound signal

A respiratory phase detection method was developed for automatic lung sound recognition without measuring the airflow. The transition points between the respiration phases and the inspiration peaks were located using the average power spectrum of lung sound. The respiratory phase pattern was evaluated from the respiration peaks to provide the time domain localization reference for lung sound feature analyzation. The method was verified using 37 recorded lung sounds. The results show that phase detection accuracy is 85.7% in the fully-automated mode and 92.3% in the semi-automated mode for normal lung sounds. Thus, the algorithm can accurately detect the respiratory phase and simplify lung research.EI0122136-21404