- 學位論文
語音訊號特徵分析在阻塞型睡眠呼吸中止症睡眠品質評量之應用
The Application of Speech Signal Characteristics Analysis for Obstructive Sleep Apnea in Sleep Quality Assessmen
摘要
傳統睡眠品質評量中的睡眠品質量表，雖然可以很快地測出結果但缺乏量化客觀數據；多重睡眠生理訊號檢測儀(Polysomnography, PSG)雖可量測多通道的生理數據但會使受測者不舒服、花費昂貴且費時，進而導致患者疏於檢查與治療。本論文利用現有的語音訊號處理之方法，主要以時域與頻域分析中得到的特徵參數將“語音訊號處理”與“睡眠品質評量”兩者結合，達到可一分鐘內提供睡眠品質評量量化數據的方法，作為輔助辨識正常人與阻塞型睡眠呼吸中止症(Obstructive-Sleep Apnea, OSA)患者。研究中比較正常人與OSA患者(患者有治療前與治療後兩組)的語音訊號，發現正常人的聲音在區域能量上的表現最穩定，OSA患者治療後則介於正常人與患者治療前之間，這表示患者經過治療後，在區域能量上確實有趨近於正常人的表現。最後本論文利用費雪線性分類法畫出一條鑑別曲線，可將正常人與OSA患者治療前分類出來，對於正常人的鑑別率有100%、對於OSA患者治療前與治療後的鑑別率都有91.7%。期望這個辨識結果可以輔助現有的睡眠品質量表來幫助醫師進行OSA患者的快速篩檢，協助醫師評估治療效果，並且可以提供患者在日常生活中簡單的自我量測當作參考，為需要的人、為醫界盡一份心力。
並列摘要
The sleep quality index of traditional sleep quality assessment is lack of digital data although the results could be obtained quickly. Polysomnography (PSG) could obtain sufficient clinical data but it is high cost, time consuming and uncomfortable for patients that discourages the patients for continuing examinations and treatment. In this study, a speech signal processing method is applied to obtain characteristic parameters from time and frequency domain analysis that combines “speech signal processing” and “sleep quality assessment” to provide the sleep quality index a quantitative data in one minute, from which the Obstructive-Sleep Apnea (OSA) patients could be diagnosed. The speech signals from normal and from OSA patients before and after treatment are compared and the results show the most stable signals in area energy are from the normal and the signal stability from OSA ones after treatment is between the normal and the OSA ones before treatment. This shows that the signal stability of OSA patients could be similar to the stability of the normal after treatment. Finally, Fisher’s Linear Discriminant is applied to draw an identification curve to classify the normal and the OSA patients before treatment, of which identification rate for the normal reaches 100% and for the OSA ones before and after treatments reaches 91.7%. It is expected that the results could serve as support for doctors to use current sleep quality index for rapid examinations and treatment effect evaluation and serve as a simple self-assessment reference for patients that is beneficial for people in need and for medical development.