Electrical signals generated by brain activity that are measured by the electroencephalogram can be distorted by electrical activity originating from eyeblinks and eye movements. This thesis proposes a new technique to identify and remove eyeblink artifacts from EEG data. An algorithm using a combination of wavelet analysis and independent component analysis (ICA) is implemented to detect the temporal location of the eyeblink artifact and eliminate it without compromising the integrity of the primary EEG data. The discrete wavelet transform is performed on 10 second epochs of data to detect the occurrence of ocular artifact. ICA is used to separate out the independent components within the data and the temporal locations of the eyeblink are used to remove the artifact and reconstruct the EEG data without that source of distortion. The results obtained indicate that the technique implemented may be robust enough to effectively process EEG data and is capable of removing eyeblink artifacts successfully when they are prominent and the data does not contain a great deal of movement artifact. The results show an 88.68% detection rate, a false positive rate of 4.03%, and an 87.23% removal rate for all eyeblinks that were accurately detected. The statistics obtained compared favorably with work done by others in this field of investigation.
Library of Congress Subject Headings
Electroencephalography--Data processing; Eye--Movements--Data processing; Wavelets (Mathematics)
Department, Program, or Center
Electrical Engineering (KGCOE)
Pesin, Jimy, "Detection and removal of eyeblink artifacts from EEG using wavelet analysis and independent component analysis" (2007). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus