Can upper airway obstruction be detected by spectral analysis of flow rate signals?

Obstructive and restrictive disorders are well recognized on flow-volume curves. However, upper airway obstructions are usually discernible only in the most extreme cases. The signal of respiratory airflow, which is measured at the airway opening, is a summation of acoustic waves (induced and reflected) of various types that depend on lung anatomy, tissue characteristics and physiological maneuver. In order to examine the hypothesis we developed a simplified in vitro model of respiratory airflow through a single airway. The airway (a 20 mm diameter Tygon tube) was connected to a Fleisch pneumotachometer at the proximal end, and to a 2 liter syringe at the distal end. Flow rates were acquired with a piezoelectric pressure transducer at very high rates (2 KHz) while the syringe was used to simulate forced breathing maneuvers. Data was collected for three tube lengths (15, 25 and 45 cm) with and without various degrees of obstruction at different locations along the tube. The respiratory airflow signal is a nonstationary signal, hence, we employed a Wigner-Ville algorithm to study the time-frequency diagrams (TFDs) of the measured data. Each test was repeated 16 times The results clearly demonstrated specific TFDs for obstructed and non-obstructed airways with a consistent variation of the TFD as the degree of the obstruction increases. The TFDs can also distinguish between distal and proximal obstructions.