# Artifact subspace reconstruction

From Brede Wiki

Topic (help)
| |
---|---|

Artifact subspace reconstruction | |

Abbreviations: | ASR |

Variations: | |

Category: | Artifact subspace reconstruction |

Parents: | |

Children: | |

Databases: | |

Search | |

Papers: | DOAJ Google Scholar PubMed |

Ontologies: | MeSH NeuroLex Wikidata Wikipedia |

Other: | Google Twitter WolframAlpha |

* Artifact subspace reconstruction* (ASR) is a processing step in EEG analysis.
In the documentation to its reference implementation in EEGLAB Christian Kothe describes it by:

- The basic principle is to first find a section of data that represents clean "reference" EEG and to compute statistics on there. Then, the function goes over the whole data in a sliding window and finds the subspaces in which there is activity that is more than a few standard deviations away from the reference EEG (this threshold is a tunable parameter). Once the function has found the bad subspaces it will treat them as missing data and reconstruct their content using a mixing matrix that was calculated on the clean data.

## [edit] Method

- 1 minute callibration data
- Principal component analysis on a sliding window
- Remove high-variance PC components 3 standard deviations above the mean
- Reconstruct the signal with the remaining