Digitala Vetenskapliga Arkivet

Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Automated Artifact Detection in Electroencephalogram Recordings
Linköping University, Department of Biomedical Engineering.
2024 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

Artifact detection in electroencephalogram (EEG) recordings is vital for ensuring data qual-ity in neuroscience research. This study evaluates the performance of an artifact detection algorithm on a 100-hour dataset, revealing significant challenges. This study uses Independent Component Analysis (ICA) to detect eye artifacts and Canonical Correlation Analysis (CCA) to identify muscle artifacts in EEG recordings in order to identify timestamps for each detected artifact. The results show that despite challenges, the algorithm achieved 92 percent accuracy in detecting ICA components. Notably, while the ICA method detects eye artifacts with a 58 percent success rate and achieves 58.9 percent in duration detection, the CCA method performs significantly better. The CCA method detects muscle artifacts at a rate of 86.7 percent and has an 84.2 percent in duration detection. Although the two methods target different artifacts, the CCA method demonstrates better detection rates compared to the ICA method when evaluated against annotated artifacts. The complexity arises from the simultaneous occurrence of various artifacts, particularly eye artifacts whose identification is hindered by their variability. The ICA- and CCA method managed to detect the majority of the artifacts. However, the CCA method exhibits promising results in detecting muscle artifacts, leveraging their distinct characteristics to enhance detection accuracy. In conclusion, while both methods are valuable methods, the CCA method gives significantly better results due to lower autocorrelation and higher frequency in muscles, making it easier for the algorithm to detect muscle artifacts.

Place, publisher, year, edition, pages
2024. , p. 46
National Category
Other Medical Engineering
Identifiers
URN: urn:nbn:se:liu:diva-206128ISRN: LIU-IMT/TFK-A-M–-24/52--SEOAI: oai:DiVA.org:liu-206128DiVA, id: diva2:1887018
Subject / course
Master's Programme in Biomedical Engineering
Supervisors
Examiners
Available from: 2024-08-07 Created: 2024-08-06 Last updated: 2024-08-07Bibliographically approved

Open Access in DiVA

No full text in DiVA

Search in DiVA

By author/editor
Lönnblad, Gustav
By organisation
Department of Biomedical Engineering
Other Medical Engineering

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 173 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf