基于ICA去除人工耳蜗EABR信号的FNS伪迹研究

Reduction of Facial Nerve Stimulation Artifacts in Electrically Evoked Auditory Brainstem Responses Based on Independent Component Analysis

为改善人工耳蜗植入者的电诱发听觉脑干反应在其人工耳蜗植入后的评估效果,针对脑电信号在叠加平均法等预处理后出现的面部神经刺激伪迹(FNS),提出了利用独立成分分析(ICA)的方法去除电诱发听觉脑干反应中的面部神经刺激伪迹,通过优化算法寻求一个分解矩阵,使目标函数最优,将预处理后的脑电信号按统计独立原则分解为若干独立信号源成分,完成伪迹的去除。对比处理前后的信号波形图,结果表明,采用ICA方法能有效地去除人工耳蜗中的FNS伪迹。

The electrically evoked auditory brainstem response （EABR） is one objective evaluation tool for co- chlear implant （CI） subjects. The EABR signals are usually obtained by averaging responses, but because of the fa- cial nerve stimulation artifacts （FNS） do not averaged out by increasing repetitions. Therefore independent compo- nent analysis is adopted in the case, which make sure the FNS artifacts be separated from the EABR. The results show that the FNS artifacts can be efficiently removed by the independent component analysis.