摘要
心冲击(BCG)伪迹抑制是同步脑电-功能磁共振成像(EEG-fMRI)时脑电伪迹抑制中最具有挑战性的任务。为此,提出一种使用深度长短期记忆(Deep-LSTM)网络的BCG伪迹抑制方法。首先使用Deep-LSTM网络拟合BCG伪迹与心电(ECG)信号之间的非线性变换,然后将ECG信号经由所拟合的非线性变换得到同步EEG-fMRI时EEG电极采集得到的混合信号中的BCG伪迹的最优估计,再从混合信号中减去BCG伪迹的最优估计,得到较高质量的EEG信号。采用实际数据进行BCG伪迹抑制实验,实验结果表明所提出的BCG伪迹抑制方法能有效抑制BCG伪迹,且在主观视觉评价、峰峰值比值和改进的归一化功率谱比值上均优于传统的BCG伪迹抑制方法。
The removal of ballistocardiogram(BCG) artifact is the greatest challenge in the electroencephalogram(EEG) artifact suppression method for simultaneous EEG-fMRI. To address this issue, a novel method is proposed to remove the BCG artifact by using the deep long short-term memory(Deep-LSTM) network. Firstly, the nonlinear transformation between BCG artifact and electrocardiogram(ECG) is estimated by using the Deep-LSTM network. Then, the optimal estimation of BCG artifact in composite signal collected by EEG electrodes during simultaneous EEG-fMRI is achieved by the ECG undergoing the estimated nonlinear transformation. Finally, the high-quality EEG is extracted by removing the optimal estimation of BCG artifact from the composite signal. Experiments on the BCG artifact removal of the real data are performed. Results show that the proposed method can successively remove the BCG artifact. It has great advantage over than the conventional BCG artifact removal methods on subjective visual evaluation, peak-to-peak value ratio and improved normalized power spectrum ratio.
作者
韩亮
黄谦
蒲秀娟
刘聪聪
柴俊杰
Han Liang;Huang Qian;Pu Xiujuan;Liu Congcong;Chai Junjie(School of Microelectronics and Communication Engineering,Chongqing University,Chongqing 400044,China;Chongqing Key Laboratory of Bio-perception&Intelligent Information Processing,Chongqing 400044,China)
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2020年第11期198-206,共9页
Chinese Journal of Scientific Instrument
基金
重庆市自然科学基金“基于非线性混合模型的心冲击伪迹抑制方法研究”(cstc2016jcyjA0376)项目资助。
关键词
心冲击伪迹
非线性变换
深度长短期记忆网络
ballistocardiogram artifact
nonlinear transformation
deep long short-term memory network
