使用鼻腔氣流訊號檢測阻塞性睡眠呼吸中止症

Detection of Obstructive Sleep Apnea Using Nasal Airflow Signal

陳嵩霖(Song-Lin Chen)

指導教授 ： 簡福榮

國立臺北科技大學/電資學院/電機工程系研究所/碩士(2014年)

https://doi.org/10.6841/NTUT.2014.00933

摘要

在睡眠障礙中，睡眠呼吸中止症（Sleep Apnea Syndrome, SAS）對人體精神及心血管方面造成不良影響，其中高達 90%以上為阻塞性睡眠呼吸中止症（Obstructive Sleep Apnea , OSA），因此檢測與治療 OSA的方式廣為學界及醫界所重視。
專業的睡眠障礙診斷是藉由在醫院的睡眠實驗室量測病患的多導睡眠圖 (Polysomnography, PSG)，包含受測者整夜睡眠過程的生理訊號。這些PSG紀錄必須經由專業睡眠技師現場操作及醫師來分析評估。因此，PSG是昂貴、費時且耗費人力的過程。
在本論文中，為了彌補PSG的缺點，發展一些出基於鼻腔氣流訊號用於OSA自動化檢測的方法替代PSG。這些方法包含振幅比較法(Amplitude Comparison Method, ACM)、特徵擷取法(Feature Extraction Method, FEM)和平均二階導數法(Mean Magnitude of the Second Derivative Method, MMSD)以及結合平均二階導數之特徵擷取法(Combined Mean Magnitude of the Second Derivative and Feature Extraction Method, MMSD-FEM)，其中平均二階導數法可以去除人為造成的雜訊及基線漂移。
實驗結果顯示，使用振幅比較法、特徵擷取法、平均二階導數法以及結合平均二階導數之特徵擷取法分別可以得到62.42%、73.76%、92.18%和78.46%的平均表現度。

關鍵字

阻塞性睡眠呼吸中止症 ； 多導睡眠圖 ； 平均二階導數法

並列摘要

Nowadays many elder people suffer from sleep disorders, especially the Sleep Apnea Syndrome (SAS). The SAS would adversely affect their cardiovascular systems and their mind. Since 90% of SAS is Obstructive Sleep Apnea (OSA), the detection and treatment of OSA has drawn the interest of medical and academic communities. The professional diagnosis of sleep disorders is to use Polysomnography (PSG) which involves overnight recordings of several physiological signals in sleep laboratories. These recorded signals are then analyzed by the sleep specialist for final diagnosis of the disease. Therefore, PSG is an expensive, time-consuming, and labor-intensive procedure.　
In this thesis, to compensate the drawback of PSG, several methods have been developed as PSG alternatives for automatic OSA detection based on the nasal airflow signal. These methods include the Amplitude Comparison Method (ACM), the Feature Extraction Method (FEM), the Mean Magnitude of the Second Derivative Method (MMSD), and the Combined Mean Magnitude of the Second Derivative and Feature Extraction Method (MMSD-FEM). The Mean Magnitude of the Second Derivative Method is able to remove the additive noise and baseline drift caused by anthropogenic factors. The experimental results show that among the ACM, FEM, MMSD and MMSD-FEM can achieve the mean performance rate at 62.42%, 73.76 %, 92.18 % and 78.46%, respectively.

並列關鍵字

Obstructive Sleep Apnea ； Polysomnography ； Mean Magnitude of the Second Derivative Method

參考文獻

[19] 陳濘宏，阻塞型睡眠呼吸中止症候群，台灣醫學9卷3期，May 2005。

[1] P. Varady, S. Bongar, and Z. Benyo, “Detection of Airway Obstructions and Sleep Apnea by Analyzing the Phase Relation of Respiration Movement Signals,” IEEE Transactions on Instrumentation and Measurement, vol. 52, no. 1, February, 2003, pp. 185-190.

[3] T. C. Huang, H. Y. Chen, and W. C. Fang, “Real-Time Obstructive Sleep Apnea Detection Based on ECG Derived Respiration Signal,” Department of Electronics Engineering and Institute of Electronics System-on-Chip Research Center, National Chiao Tung University, Taiwan (R.O.C.), 2012, pp. 341-344.

[4] Electrocardiography, http://en.wikipedia.org , June , 2014.

[5] J. W. Hurst, “Naming of the Waves in the ECG, With a Brief Account of Their Genesis ,” Circulation.98, 1998 , pp. 1937-1942.