應用可攜式腦波機於睡眠呼吸阻斷 與良導絡相關性之研究

Abstract

本篇論文主要是利用一無線數位化腦波檢測系統，進行睡眠呼吸阻斷症研究與腦波及良導絡相關性研究。此腦波擷取系統使用藍芽(Bluetooth)晶片模組與具有省電且功能強大的MSP430微控制器(Microcontroller, MCU)，整合前置放大器、濾波器、增益放大器、及數位控制電路研製而成。本裝置能迅速地擷取腦波訊號于以數位化並將之透過藍芽模組傳至PC端。PC主機可將病患之已數位化的腦電波訊號，以NAB (Non-linear energy operator, AR model, and Bisecting k-means algorithm)方法與Bisecting k-means algorithm方法進行分類分析，並可將長時段腦波訊號資料分類與儲存。接著，應用小波轉換(Wavelet transform)擷取阻塞型睡眠呼吸中止症候群(Obstructive Sleep Apnea Syndrome, OSAS)所引發的腦波訊號特徵，再經由自行設計的局部特徵分析(Characteristic part analysis, CPA) 之類神經網路架構等進行學習訓練，並分析及判讀睡眠呼吸阻斷的發生過程。經系統統計與評估結果，本系統的辨識成效最高可達Sensitivity約69.64％，Specificity約44.44％，已具有臨床參考的價值。期能提供醫療專業人員作為輔助診斷的工具，進而能提升醫療服務效率。 除此之外，本研究亦藉由良導絡值與腦波訊號節律的量測，探討人類腦部活動分別處於Alpha 波、Beta波期間，與身體十二條經絡的良導絡值之關聯性。經實驗結果證實腦部活動處於不同的腦波期間，經絡的良導絡值會有明顯的差異存在，將可供醫療專業人員參考。 關鍵字：類神經網路、藍芽、腦電訊號、良導絡、阻塞型睡眠呼吸中止症候群、小波轉換。

In this dissertation, a digital wireless electroencephalograph (EEG) acquisition and recording system was adopted to analyze the obstructive sleep apnea syndrome and investigate the relation between the EEG signal rhythms and Ryodoraku. The EEG acquisition and recording system uses a Bluetooth chip module and an energy-saving MSP430 Microcontroller (MCU) with powerful functions, along with integrated pre-amplifiers, filters, gain amplifiers, and a digital control circuit. After quickly acquiring and digitizing EEG signals, this system transfers the signals to a PC via the Bluetooth module. The PC then uses the NAB (Non-linear energy operator, AR model, and Bisecting k-means algorithm) method and bisecting k-means algorithm to classify and analyze patients' EEG signals. The system first performs long-period EEG signal classification and storage, and then applies wavelet transforms to acquire EEG signal characteristics due to obstructive sleep apnea syndrome (OSAS). We trained a characteristic part analysis (CPA) artificial neural network designed by our group so that it could analyze and interpret the occurrence of OSAS, and compile and assess data. The system had a maximum sensitivity of approximately 69.64%, and a specificity of approximately 44.44%. This indicates that the system can provide clinical medical personnel with a valuable auxiliary diagnostic tool, improving medical service efficiency. In addition, this research study the correlation analysis of EEG signal rhythms and Ryodoraku value of 12 acupuncture meridian of human body in Alpha wave and Beta wave brain activity periods, respectively. The experimental results have been confirmed that the Ryodoraku value had obvious difference in different brain wave periods which could provide reference for clinical medical personnel. Keywords: Artificial Neural Network, Bluetooth, EEG, Ryodoraku, Obstructive Sleep Apnea Syndrome, Wavelet Transform.