Neural Networks for Pattern Recognition
Neural Networks for Pattern Recognition
On the behavior of information theoretic criteria for model orderselection
IEEE Transactions on Signal Processing
Extracting multisource brain activity from a single electromagnetic channel
Artificial Intelligence in Medicine
Hi-index | 0.00 |
Electroencephalogram (EEG) recordings are often distorted by high-amplitude artifacts which hamper its correct visual inspection. In this work we present a method which can be applied separately to each channel to extract high-amplitude components. The method is called local singular spectrum analysis (SSA) and is a principal component analysis in clusters formed after embedding the signals in their time-delayed coordinates. The extracted signal can be subtracted from the original channel resulting in a corrected EEG version. The algorithm is applied to real EEG segments containing paroxysmal epileptiform activity contaminated by artifactual activity. The extracted artifact as well as the corrected EEG will be presented.