多通道腦電波量測系統之研製

Abstract

腦機介面(BCI, Brain Computer Interface)提供一個全新的通訊方式，使因神經肌肉受損或患有肌肉萎縮性脊髓側索硬化症等病患，可以透過腦部的活動直接控制外部的裝置，而不需經過神經及肌肉的媒介。但是腦機介面的相關設備，不僅昂貴且體積龐大，對於腦機介面的普及，實不是一項優勢。 因此，本研究目的在於研製一多通道、低成本、小體積及模組化之腦電波擷取系統。本系統包含基本的電子電路，可調適增益放大器、交換式電容濾波模組及以LabVIEW為平台所設計之應用軟體。 硬體設計上，以驅動右腳線路消除共模雜訊，以單晶式交換式電容濾波器，取代傳統主動RC濾波器降低頻率偏移，以數位電阻器取代傳統可變電阻，提高精確度，去除人為調整造成之誤差。印刷電路板佈局上，加入模組化的概念，使腦電波放大電路可進行堆疊，擴充通道數。 製作完成後，將本系統與腦電波專業量測儀器NuAmps進行比較。結果顯示，誤差評估在時域上的誤差平均值與標準差分別為0.1809伏，及0.1153；在頻域上誤差平均值與標準差分別為0.0606伏及0.167。趨勢相似度評估的相似度平均為83.25%。相關評估的相關係數可達0.8以上。 根據實驗結果，說明所研製之多通道腦電波量測系統具可行性。未來期望能將本系統與嵌入式系統做結合，可協助行動不便之使用者獲得更好的生活品質。Brain Computer Interface (BCI) provides a novel communication method for patients with neuromuscular disorders or amyotrophic lateral sclerosis to directly control external devices with their brain activities, without pass conventional motor output pathways of nerves and muscles. However, the BCI relevant equipment is expensive and bulky, it is hard to be popularized. The objective of this research is to design and implement a multi-channel, low cost, small-volume and modularized Electro-Encephalo-Gram (EEG) acquisition system. The system consists of fundamental circuits, the programmable gain amplifiers, the switch capacity filter modules and the LabVIEW-based application software. Concerning the design of hardware, the Drive Right Leg (DRL) is adopted to reduce the common-mode noise. The switch capacity filter is used to replace the traditional active RC filter to reduce the drift of frequency. The digital potentiometer is used to replace traditional rheostat to increase the accuracy and reduce the error caused by artificial adjustment. The concept of module is introduced into the layout of Printed Circuit Board (PCB) to increase the numbers of channel through circuit stacking. After the system is developed, the measurement performance of the system is compared with the NuAmp which is a professional instrument of BCI. The results show that the average error and the standard deviation are 0.1809 volts and 0.1153, respectively. In frequency domain, the average error and the standard deviation are 0.0606 volts and 0.167, respectively. The average of trend similarity is 83.25%. The correlation coefficient is larger than 0.8. According to the experiment results, the system is feasible. In the future, the system can be integrated with an embedded system for benefits disabled persons.