Abstract
The prevalence of sleep apnea syndrome (SAS) is estimated about 2% of the population in Japan. SAS is generally diagnosed using polysomnography (PSG) in the clinic. On the other hand, the cost and inconvenience of PSG make simplified technique desirable for diagnosing SAS. Accordingly, screening test for SAS using pulse oximetry is often performed at home. However, the motion artifact is known to affect the accuracy and reliability of pulse oximetry, and the sensor attachment have a negative influence on sleeping condition. It is therefore necessary to prevent harmful effects on sleeping condition. To address this problem, we have investigated the non-contact measurement of blood flow using the Microsoft KinectTM sensor. Given streams of RGB color images, the data of time series variation have been calculated from these subtracted images. Preliminary comparison experiments between the pulse rate values obtained by these data and by the photo-plethysmography showed a fairy good agreement. Further analysis of this data indicated that the data obtained from the Microsoft KinectTM sensor might be considered a probability of breathing waveform.
