智慧化臨床資訊自動篩選與分析技術研究

The study of intelligent auto-screening and analysis techniques in clinical information

溫國棟(Kuo-Tung Wen)

指導教授 ： 蔡育秀

中原大學/工學院/醫學工程研究所/碩士(2002年)

https://doi.org/10.6840/cycu200200526

摘要

睡眠窒息症(Sleep Apnea Syndrome)是常見的睡眠疾病，會令患者在睡眠中出現頻繁的呼吸暫停，因而影響健康，甚至可能導致死亡。目前，國內至少有二十萬人深受睡眠窒息症的困擾，並且有逐年增加的趨勢，因而造成醫護人員，判讀受測者是否罹患睡眠窒息症的工作負擔日益加重。故如何協助醫護人員，提升睡眠窒息症的判讀成效，再加以適當的治療，將是一個重要的課題。口鼻腔氣流、血氧飽和度、胸部腹部呼吸運動和心電圖等，都是近年來臨床研究用來判讀睡眠窒息症的生理訊號。本研究則針對尚未受到重視，也能用來判讀睡眠窒息症的腦波圖(EEG)，研發睡眠窒息症之診斷分析法則，並提供醫護人員具參考價值的腦波圖判讀結果。
　　本研究以小波轉換(Wavelet Transform)來發掘隱藏在睡眠窒息症內的腦波特徵，腦波訊號經過觀察分析後，發現睡眠窒息停止時，腦波會發生Theta波、Alpha波頻帶轉移的現象，亦即可當作睡眠窒息症之腦波特徵，再經由統計方法研發睡眠窒息症之診斷分析法則。
　　為了評估自行研發的腦波圖診斷分析法則之可行性，採判定十位受測者是否罹患睡眠窒息症來驗證。系統判讀成效sensitivity約54%以及specificity約56%，已達到臨床參考價值的標準，故醫護人員可藉由腦波圖的判讀結果，作為決策支援，進而提升整體的醫療品質。

關鍵字

小波轉換 ； 睡眠窒息症 ； 腦波圖

並列摘要

Sleep Apnea Syndrome is a common sleeping disorder and is considered as the repeat obstruction of airflow during sleep. Long term Sleep Apnea would affect health condition, or even result in death. Approximately, there is at least 1% of the population in Taiwan suffered from moderate to severe sleep apnea, and the percentage has increased in the recent years. To diagnose sleep apnea thoroughly would cause a heavy workload, since it requires long period monitoring. How to increase the efficacy of diagnosing sleep apnea with modern computer technology became an important issue. Recently, there are a few physiological signals used to diagnose sleep apnea, such as respiratory effort, airflow, blood oxygen levels, and heart rate. This research integrated the EEG analysis to develop a diagnosis algorithm for the sleep apnea detection. The study uses the wavelet transform to extract the brainwave feature from EEG signal. It is found that, right after the sleep apnea is terminated, the theta and alpha wave have abrupt frequency shift. This phenomenon is used as a criterion to identify sleep apnea in this study.

　In order to evaluate the feasibility of the self-developing diagnosis, 10 sets of EEG signal were used to test the algorithm. The results exhibit a sensitivity of 54% and specificity of 56%. It meets to the clinical application requirement, and can be used as an auxiliary decision support for long term sleep apnea detection.

並列關鍵字

EEG ； wavelet transform ； Sleep Apnea Syndrome

參考文獻

1.M Varanini, Apnea Detection from 24-Hour Recordings of Respiration in Chronic Heart Failure Patients, Computers in Cardiology 27, pp. 489-492, 2000.

2.JE Mietus, Detection of Obstructive Sleep Apnea from Cardiac Interbeat Interval Time Series, Computers in Cardiology 27, pp.753-756, 2000.

3.Péter Várady, Detection of Airway Obstruction and Sleep Apnea by Analyzing the Phase Relation of Respiration Movement Signals, IEEE Instrumentation and Measurement Technology Conference, pp. 185-190, 2001.

4.Péter Várady, On-line Detection of Sleep Apnea During Critical Care Monitoring, Proceedings of the 22nd Annual EMBS International Conference, pp. 1299-1301, 2000.

5.Martin Kermit, Treatment of obstructive sleep apnea syndrome by monitoring patients airflow signals, Pattern Recognition Letters 21, pp. 277-281, 2000.