A robust detection algorithm to identify breathing peaks in respiration signals from spontaneously breathing subjects

Assessing respiratory and cardiovascular system coupling can provide new insights into disease progression, but requires accurate analysis of each signal. Respiratory waveform data collected during spontaneous breathing are noisy and respiration rates from long term physiological experiments can vary over a wide range across time. There is a need for automatic and robust algorithms to detect breathing peaks in respiration signals for assessment of the coupling between the respiratory and cardiovascular systems. We developed an automatic algorithm to detect breathing peaks from a respiration signal. The algorithm was tested on respiration signals collected during hemorrhage in a conscious ovine model (N=9, total length = 11.0h). The breathing rate varied from 15 to as high as 160 breaths/min for some animals during the hemorrhage protocol. The sensitivity of the algorithm to detect respiration peaks was 93.7% with a precision of 94.5%. The developed algorithm presents a promising approach to detect breathing peaks in respiration signals from spontaneously breathing subjects. The algorithm was able to consistently identify breathing peaks while the breathing rate varied from 15 to 160 breaths/min.