Recovering EEG brain signals: Artifact suppression with wavelet enhanced independent component analysis
Recovering EEG brain signals: Artifact suppression with wavelet enhanced independent component analysis
复制标题
DOI：
10.1016/j.jneumeth.2006.05.033
复制
发表时间：
2006-12-15
影响因子：
3
通讯作者：
Makarov, Valeri A.
中科院分区：
文献类型：
--
作者：
Castellanos, Nazareth P.;Makarov, Valeri A.
独立成分分析 (ICA) 已被证明对于抑制脑电图记录中的伪影非常有用。它涉及将测量信号分离成统计上独立的分量或源，然后拒绝那些被认为是人为的信号。我们表明，感兴趣的大脑活动“泄漏”到标记为人工的成分中意味着人们将失去该活动。为了克服这个问题，我们提出了一种新的小波增强 ICA 方法 (wICA)，该方法不将小波阈值应用于观察到的原始 EEG，而是将其作为中间步骤应用于分离的独立分量。它允许恢复“人造”组件中存在的神经活动。采用半模拟和真实脑电图记录，我们量化了时域和频域中 ICA 和 wICA 伪影抑制引入的脑电图大脑部分的失真。在研究皮质电路的背景下，我们还评估了 ICA/wICA 校正脑电图的频谱和部分频谱相干性。我们的结果表明 ICA 可能会导致神经功率谱的低估和不同皮质位点之间的一致性的高估。 wICA 伪影抑制保留了潜在神经活动的频谱（幅度）和相干性（相位）特征。 (c) 2006 Elsevier B.V. 保留所有权利。
Independent component analysis (ICA) has been proven useful for suppression of artifacts in EEG recordings. It involves separation of measured signals into statistically independent components or sources, followed by rejection of those deemed artificial. We show that a "leak" of cerebral activity of interest into components marked as artificial means that one is going to lost that activity. To overcome this problem we propose a novel wavelet enhanced ICA method (wICA) that applies a wavelet thresholding not to the observed raw EEG but to the demixed independent components as an intermediate step. It allows recovering the neural activity present in "artificial" components. Employing semi-simulated and real EEG recordings we quantify the distortions of the cerebral part of EEGs introduced by the ICA and wICA artifact suppressions in the time and frequency domains. In the context of studying cortical circuitry we also evaluate spectral and partial spectral coherences over ICA/wICA-corrected EEGs. Our results suggest that ICA may lead to an underestimation of the neural power spectrum and to an overestimation of the coherence between different cortical sites. wICA artifact suppression preserves both spectral (amplitude) and coherence (phase) characteristics of the underlying neural activity. (c) 2006 Elsevier B.V. All rights reserved.