Fourier analysis of airflow and transpulmonary pressure signals filtered on-line.

Airflow and transpulmonary pressure curves obtained from a normal man and a dog, during hyperventilation or when breathing spontaneously, were analysed on a digital computer by means of the fast Fourier transform. The same curves were also analysed after passing through an on-line RC filtering algorithm having cutoff frequencies extending from 10 down to 0.1562 Hz. Results show that respiratory parameters can contain a large amount of biological background noise mainly related to cardiovascular motion. The only instance in which the magnitude of fundamental harmonic was not the largest was for the airflow of the dog measured during spontaneous breathing. It was concluded that the cutoff frequency should be adjusted to a value 3-4 times that of the fundamental frequency. More pronounced filtering would cause important distortions of the physiologic signals.