基于扩展Infomax与DOST的多导联睡眠脑电伪迹去除方法研究

发布时间：2018-06-18 11:48

本文选题：脑电信号 + 睡眠　；参考：《计算机应用研究》2017年12期

【摘要】：对多导联脑电信号(EEG)的预处理对于睡眠疾病的检测、预防和治疗有着重要的意义,传统的S变换法在处理多导联EEG信号时在伪迹自动识别、计算复杂度和阈值选取这三方面存有不足,影响了该方法的应用效果。为了提高该方法的应用效果,采用一种扩展Infomax与离散正交S变换结合的伪迹去除方法(DOSTIN),根据分形维数法实现伪迹识别,利用最大类间方差法使阈值的选取避免了过度去噪造成的有用信号损失。采用麻省理工的公开数据集验证去噪后的效果,结果表明DOSTIN法在保证去噪效果的同时,比ICA与S变换相结合方法的计算复杂度降低了30%,证明该方法在多导联信号去噪方面具备可行性。
[Abstract]:The preprocessing of multilead EEG signal (EGG) plays an important role in the detection, prevention and treatment of sleep diseases. The traditional S-transform method is used to automatically recognize artifacts in the processing of multi-lead EEG signals. The computational complexity and threshold selection are insufficient, which affects the application effect of this method. In order to improve the application effect of this method, an extended Infomax method combined with discrete orthogonal S transform is used to remove artifacts. According to fractal dimension method, artifact recognition is realized. The maximum inter-class variance method is used to avoid the loss of useful signal caused by excessive denoising. The effect of de-noising is verified by using the open data set of MIT. The results show that the DOSTIN method can ensure the denoising effect at the same time. Compared with the method of ICA and S-transform, the computational complexity is reduced by 30%. It is proved that this method is feasible in multi-lead signal denoising.
【作者单位】：重庆邮电大学通信与信息工程学院无线传输重点实验室;重庆金瓯科技发展有限责任公司;
【基金】：重庆市重点产业共性关键技术创新专项资助项目(cstc2015zdcy-ztzx40008)
【分类号】：TN911.7

【相似文献】