Department of Psychology, Goldsmiths, University of London, New Cross, London, UK.
Psychophysiology. 2010 Sep;47(5):955-60. doi: 10.1111/j.1469-8986.2010.00995.x. Epub 2010 Mar 22.
Blink-related ocular activity is a major source of artifacts in electroencephalogram (EEG) data. Independent component analysis (ICA) is a well-known technique for the correction of such ocular artifacts, but one of the limitations of ICA is that the ICs selected for removal contain not only ocular activity but also some EEG activity. Straightforward removal of these ICs might, therefore, lead to a loss of EEG data. In this article a method is proposed to separate blink-related ocular activity from actual EEG by combining ICA with a novel technique, empirical mode decomposition. This combination of two techniques allows for maximizing the retention of EEG data and the selective removal of the eyeblink artifact. The performance of the proposed method is demonstrated with simulated and real data.
眨眼相关的眼动是脑电图（EEG）数据中主要的伪迹来源。独立成分分析（ICA）是一种用于校正这种眼伪迹的知名技术，但 ICA 的一个局限性是，选择去除的 IC 不仅包含眼动，还包含一些 EEG 活动。因此，直接去除这些 IC 可能会导致 EEG 数据丢失。本文提出了一种将 ICA 与一种新的技术——经验模态分解（EMD）相结合，从实际 EEG 中分离出与眨眼相关的眼动的方法。这两种技术的结合可以最大限度地保留 EEG 数据，并选择性地去除眨眼伪迹。通过模拟和真实数据验证了所提出方法的性能。
