Generation of optimum pseudorandom signals for respiratory impedance measurements

Spontaneous breathing may impair the reliability of forced oscillatory impedance estimates at low frequencies, especially when the oscillatory power is distributed among many frequency values. Since the amplitude of the external forcing is limited to avoid non-linearities, it is suggested that the total energy of a composite electrical signal driving the loudspeaker be maximized at a given amplitude by finding the optimum phase relationships of the signal components, and that the low-frequency components increase in energy at the expense of the less disturbed high-frequency region. In healthy children and adults and in obstructed patients, the coherences and the coefficients of variation of the respiratory system impedance (Zrs) at 2 and 3 Hz were studied in the case of three test signals of 2-15 Hz bandwidth. Signals T1 and T2 had a flat power spectrum, whereas the components of T3 decreased sharply between 2 and 5 Hz; T1 was generated by simple random selection of phase angles, while optimization for maximum energy was done for T2 and T3. Optimization alone (T2) increased the reliability of the Zrs estimates at all frequencies, whereas enhancement of the low-frequency power (T3) resulted in a radical improvement of the estimates at 2 and 3 Hz, without loss in reliability at higher frequencies.