Original Article

Automatized classification of sleep apnea syndrome using upper airway temperature change and oxygen saturation signals

  • Doğan D. Demirgüneş
  • Gökhan Ertaş
  • Timur Akçam
  • Hakan Genç
  • Ziya Telatar
  • Osman Eroğul
  • Fuat Özgen
  • Hayati Bilgiç

Received Date: 08.03.2011 Accepted Date: 06.05.2011 Gulhane Med J 2011;53(3):182-188

Snoring is a common complaint in society and the earliest and the most common symptom of sleep apnea syndrome. Although the gold standard for the diagnosis of sleep apnea is polysomnography performed under observation, there is a need for more simple diagnostic methods and screening tests. In this study, simultaneous upper airway temperature changes and oxygen saturation signals were recorded in 46 patients during one-night stay at sleep laboratory using a special signal acquisition setup. Polysomnography recordings were also obtained in the same patients. No significant differences were observed between the apnea, hypopnea and apnea-hypopnea indices extracted from the autoregressive model computed for the acquired signals and the values of these indices determined by using polysomnography (p>0.05). High linear correlations are present for all indices (r=0.94-0.99, p<0.001). Apnea index and apnea-hypopnea index, and hypopnea index extracted were approximately 10% and 7% higher than the ones determined using polysomnography, respectively. In differential diagnosis of sleep apnea, specificity, sensitivity, positive predictive value, negative predictive value and accuracy were 85%, 89%, 88%, 85% and 87% in a threshold of apnea-hypopnea index for 15 events/hour in almost equally-sized patient groups. Automatic classification of sleep apnea in accordance with the results of polysomnography is possible using the designed signal acquisition setup and the proposed signal processing technique.

Keywords: Oxygen saturation, sleep apnea, upper airway temperature change