基于视觉诱发多事件相关电位的脑-机接口系统研究

Brain-computer interface (BCI) is a frontier interdisciplinary technique based on computer science, pattern recognition and biomedical engineering, which could be widely used in real life for patients or healthy people. Recently, the research on visual evoked event related potential-based BCI has been a hot topic. High-quality ERPs could be obtained by optimizing the method to evoke ERPs and the online performance of BCI could be improved by online learning methods. Furthermore, machine vision was used to improve the performance of BCI, when it was used in practical events. In this project, we will survey the methods to evoke multi-event related potentials (mERPs) and test the recognizability of each ERP. The goal of this work is to evoke reliable ERPs for BCI system which will be improved in accuracy and can be used by different users. Online updating method for electrodes selection and online learning mode for classification will be developed, which will improve the online performance of BCI system and can be used to classify some ERPs which are changing with repetition stimuli. Machine vision will be used in BCIs, which will improve the flexibility of BCI system in practical application. The goal of this project is to improve the performance of BCI system and accelerate the pace of its application in real life.

脑-机接口技术是计算机科学、模式识别和生物医学等多学科交叉的前沿领域，有着广泛的实际应用背景。而基于视觉诱发ERP的方法是脑-机接口领域的研究热点，它可以为脑-机接口提供更稳定、准确的信号信息，并且通过在线识别算法来改进ERP脑-机接口的在线性能，再利用机器视觉方法有效提高ERP脑-机接口的实用性。本课题首先通过多ERP的诱发机制研究，通过大量实验和多ERP成分的可分性分析，设计准确性高，稳定性好的多ERP诱发范式。然后，通过在线电极选择和在线分类学习算法的研究来实现具有在线更新功能的脑-机接口系统，提高脑-机接口系统的在线稳定性，解决部分ERP随刺激时间变化的问题。最后，把脑-机接口系统与机器视觉相结合，通过对不同目标和不同环境的大量实验结果，得到一个命令简捷，使用灵活的脑-机接口系统，提高脑-机接口系统的实用性。通过本课题的研究将有效提高脑-机接口的性能，加快脑-机接口的实用化进程。

本项目按计划先后从事件相关电位 (Event-related potentials, ERP) 的诱发机制及可分性，多ERP的视觉诱发模式优化，多ERP的脑-机接口在线学习算法研究和多ERP脑-机接口系统与机器视觉的融合等几个方面展开研究。首先通过新颖的刺激范式探索有效的多ERP诱发方法。利用人脸，非匹配模式等多种方式提高所诱发的多ERP的可识别性。通过表情，卡通人脸和错误规避编码方式优化脑-机接口的诱发范式进一步提高诱发信号的可识别性，并且降低诱发范式给使用者带来的疲劳度。通过P300电位和稳态视觉诱发电位(steady state visual evoked potentials)的融合方法研究提高融合系统的性能和拓展视觉诱发脑-机接口系统的人群适用性。提出电极选择，小样本训练和在线学习模型，提高视觉诱发脑-机接口系统的实用性。最后，建构机器视觉与脑-机接口的融合演示系统，实现视听觉互补的优化系统。总体来说，本项目圆满的完成了项目提出的研究目标，今后将基于本项目进一步拓展研究，取得更多优秀成果。