期刊
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
卷 38, 期 -, 页码 224-235出版社
ELSEVIER SCI LTD
DOI: 10.1016/j.bspc.2017.06.012
关键词
Electroencephalography; Artifact detection; Machine learning; Brain signal processing
资金
- National Science and Engineering Research Council (NSERC) of Canada
Recent developments in electroencephalography (EEG) have led to a variety of consumer-grade EEG devices for brain-computer interfacing (BCI) and neurofeedback (NF) equipped with only one or a few EEG sensors. With minimal electrode coverage, most methods of detecting artifacts in the signal which arise from non-brain sources are not applicable. Furthermore, methods which can be used on single channel EEG are typically not sufficiently accurate or fast for BCI and NF applications. In this paper a new highly accurate artifact rejection method is introduced, called Filter-Bank Artifact Rejection (FBAR), which is designed for real-time EEG applications using just a few or even a single EEG channel. FBAR is compared to a current state-of-the-art method, Fully Automated Statistical Thresholding for EEG artifact Rejection (FASTER). FBAR outperformed FASTER on all test data, due mainly to its ability to detect small artifacts in the presence of high amplitude EEG. This makes FBAR particularly useful for BCI and NF applications, which are especially dependent on achieving the highest possible signal-to-noise ratio in a real-time setting. A MATLAB toolbox allowing for use and experimentation with FBAR including several customizable options is available as a Git repository at https://bitbucket.org/kiretd/FBAR. (C) 2017 Elsevier Ltd. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据