The BCI Application of Visual Evoked Potential

• Main Authors: Hong, Wei-Jhe, 洪煒哲
• Other Authors: Chen, Shih-Chung
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/44352885307882609878

碩士 === 南台科技大學 === 電機工程系 === 98 === EEG (Electroencephalogram, EEG) is an important tool used to record and observe the changes in brain activity. The clinical and neuroscience research community has been heading up research in this field. Brain Computer Interface (BCI), makes some people independent on peripheral nerves and muscles and is a novel communication and control technology. It is designed to help patients suffering from motor neuron disease (Amyotrophic Lateral Sclerosis) or patients suffering from an aftermath of major accidents leading to serious physical impairment. Using the brain and some external devices a link can be established to achieve communication with the outside world and let the patient gain control on some elements in their surrounding environment. In this study, LabVIEW software was used as the core of the BCI system. It made use of the world-famous NuAmps created by NeuroScan Company. In the process, EEG signals are retrieved simultaneously, and LabVIEW core signal processing was used to perform further mutual authentication. Off-line analysis in the BCI system utilized visual stimulation to induce the individual visual evoked potential(VEP) and adopted independent component analysis(ICA) method to resolve the interference caused by the background noise. In addition, EEG features are then extracted after off-line analysis of the data; EEG feature are then used for real-time BCI system. LEGO bricks were used to create the physical component of the page turner machine that uses the EEG signals to control the external environment. Results of these experiments taken from seventeen subjects through offline and real-time analysis of the BCI control system showed a recognition accuracy rate of 93.6% and average time around 4.95 seconds for controling the page turner robot. Regarding the bed control experiment, the results also showed a recognition accuracy rate of 88% and average time around 8.6 seconds.