Abstract: Objective To develop a steady state visual evoked potential-based brain computer interface communication system (SSVEP-BCI-CS), and improve the quality of life of patients with poststroke language disorder. Methods The theoretical framework of SSVEP-BCI-CS was designed, according to the clinical characteristics and needs of patients with post-stroke language disorder. 18 healthy volunteers were selected as the subjects. The system performance such as information transfer speed, accuracy and

recognition length were evaluated. National Aeronautics and Space Administration-task load index

(NASA-TLX) was used to evaluate the feasibility of this system. The gender differences in BCI performance and task load index were analyzed. Results SSVEP-BCI-CS was developed, including three modes of emergency call, instruction and typing. Occipital EEG were collected and decoded by spatio-temporal equalization dynamic window algorithm, with voice broadcast and screen feedback functions. The verification results show that the average information transfer speed of typing single task was 79.92±43.55 bits/min, and the accuracy was 94.00% (92.00%-97.00%); in the typing continuous task, the median output speed of Chinese characters was 1.79 (1.49-3.05) words/min; in instruction single task, the median output speed was 32.44 (11.73-38.64) instructions/min, the median accuracy was 95.65% (90.76%- 98.37%); in instruction continuous task, the average output speed was 23.66±9.50 instructions/min. The performance indexes of females were better than that of males. Conclusions All subjects can use this system to achieve the task of emergency call, instruction output and Chinese-English spelling. The system is expected to improve communication efficiency in patients with post-stroke language disorder.

Key words: Steady-state visual evoked potential; Brain computer interface; Stroke; Language disorder

摘要： 目的 研发基于稳态视觉诱发电位脑机接口的卒中后语言障碍辅助交流系统，即卒中语言之星系 统，以提高卒中后语言障碍患者的生活质量。 方法 根据卒中后语言障碍患者的临床特点和应用需求，设计卒中语言之星系统的理论构架。统计 18名健康受试者完成指令及打字任务的系统性能指标（如信息传输速率、准确率、识别时长）评估系 统可行性，采用美国国家航空航天局任务负荷指数量表评估使用负荷，并分析比较男性与女性受试 者的性能评价指标及使用负荷评价指标。 结果 卒中语言之星系统包含紧急呼叫、指令与打字三种模式，通过采集受试者枕叶脑电信 号，采用空时均衡动态窗算法解码，具有语音播报和屏幕反馈功能。验证结果显示，受试者利用 卒中语言之星系统完成打字-单任务平均信息传输速率为79.92±43.55 bits/min，准确率为94.00% （92.00%～97.00%）；古诗任务输出速率为1.79（1.49～3.05）字/分钟；指令-单任务输出速率为32.44 （11.73～38.64）指令/分钟，正确率为95.65%（90.76%～98.37%）；指令-连续性任务输出速率为 23.66±9.50指令/分钟。女性受试者打字-单任务中平均信息传输速率、正确率，指令-单任务中输出 速率和正确率，指令-连续性任务中输出速率均高于男性受试者，差异均有统计学意义。 结论 受试者均能借助所构建系统实现紧急呼叫、指令输出和汉英拼写，该系统有望在临床环境中 提升沟通的效率。

关键词: 稳态视觉诱发电位; 脑机接口; 卒中; 语言障碍