標題: 以穩態視覺誘發電位為基礎之無線腦機介面進食協助系統
A Wireless SSVEP-based BCI Eating Assistive System
作者: 邱晴瑜
林進燈
Chiu, Ching-Yu
Lin ,Chin-Teng
影像與生醫光電研究所
關鍵字: 腦機介面;醫療照護;穩態視覺誘發電位;Brain computer interface;healthcare;SSVEP
公開日期: 2016
摘要: 醫療照護系統一直是腦機介面應用的重要發展方向之一。由於腦機介面旨在讓使用者能夠透過腦波自主照護,對於減輕照護者的負擔、被照護者心理壓力或者因應全球性人口老化所致、不斷攀升的醫療照護需求,腦機介面照護系統的發展都是值得投入的、可行的解決方案。 目前被提出的腦機介面照護系統多半有如下的缺點: 使用配備大量線材的濕式電極,使用前需耗費相當的準備時間,實用上也不方便;提供的功能較為單一,大多為拼字、或者撥打電話,照護者往往需要在旁全程協助,照護與被照護雙方的身心負擔依舊。 本論文提出一套以穩態視覺誘發電位為基礎之無線腦機介面進食協助系統。由於採用無線乾式電極腦波帽,使用上極為便利。功能上,本系統可供使用者以腦波操控機器手臂自主進食,並嘗試性地結合影音播放、視訊電話、主動互動等功能,除了減輕照護者的工作負擔之外,也提升了使用者的使用體驗。為提升系統可使用性,本論文亦針對此系統適用之演算法加以改良,以減低系統之錯誤率。根據分析結果,本系統之平均準確度達90.84%,平均訊息傳輸率為20.36(bit/min)。實用性測試上也邀請行動不便的病友參與,並獲得基本之肯定,與進一步改善之建議。
Healthcare system has been one of the important development directions of Brain Computer Interfaces (BCI) applications. Because BCI aims at enabling users to perform tasks through their brain waves, it is a feasible and worth developing solution to problems such as physiological and psychological burdens of caregivers and care receivers and the growing healthcare demand due to global aging. Current proposed BCI systems are often with the following disadvantages: equipped with wet sensors requiring lots of wires and lots of preparation, hence lowering the applicability of the system; providing single functions such as spelling or calling which still leaves users requiring full assistance from the caregivers, hence doesn’t provide much help for reducing their burdens. This work proposed a wireless SSVEP- based BCI eating assistive system. Adoption of a wireless EEG cap, Mindo, with dry sensors makes the system easy to use. Functionally, this system not only enables users to control a robotic arm to have their meals independently through their brain waves but also tries to integrate functions such as video entertainment, video call, and active interaction, hence it can both reduce burdens of caregivers and improve the user experience. To enhance usability, this work also developed an algorithm fitting this system to lower error rate. This system achieved an average accuracy of 90.84%, and ITR of 20.36 bit per min. For usability testing, this work also invited an user with physical disabilities to participate in. The user gave basic affirmation and provided suggestions for further improvement.
URI: http://etd.lib.nctu.edu.tw/cdrfb3/record/nctu/#GT070258233
http://hdl.handle.net/11536/143193
顯示於類別:畢業論文