Automated detection of respiratory phases by acoustical means

Detection of respiratory phases by airflow measurement has been used in swallowing and respiration studies. However, applying the traditional airflow measurement techniques in studies on children with significant neurological impairments is very difficult and may hamper the entire assessment. This paper proposes a noninvasive method based on the use of respiratory sounds as an alternative technique of airflow measurement. Respiratory sounds were recorded from 11 healthy children and 6 healthy adults with accelerometers placed over trachea and five locations of the anterior chest. The average power spectra of known inspiration and expiration were calculated for each signal. The best recording site on the chest was determined by the greatest difference in average power between respiratory phases. The best recording site was found on the left side for 69% and on the right side for 31% of our subjects´ population. A custom written program used the sound signal from trachea and the best site on the chest to detect the respiratory phases. The results were compared with airflow, which was measured simultaneous to the breath sounds, and showed a reliable match for the detected phases