Master of Science
Electroencephalogram (EEG) is a widely used technique to record electrical brain activity. It is prone to be contaminated by non-neuronal sources that can generate artifacts in the signal due to its sensitivity and its poor signal-to-noise ratio. One of the main challenges in analyzing EEG data is the systematical and effective removal of artifacts from the signal. Although many methods have already been introduced to approach this issue, there is still no robust method for handling all sources of contaminations. For example, eye blinking is a physiological artifact occurring very frequently in spontaneous EEG recordings and therefore, removing these artifacts in a systematic way is a compelling need. The aim of this research is to build an automated pipeline to detect eye blinking artifacts in EEG signals using the generalized Ising model to act as a pattern recognition algorithm. A sample blink pattern is extracted from a single subject whose blink events are validated and marked by an EEG expert. The generalized Ising Model Algorithm works as a fully automated method for identifying all epochs similar to the eye blink pattern. Using the proposed method to discriminate the blinks artifact in continuous EEG data yields optimistic results. From eight healthy subjects, the results show high level of accuracy (90.5 %).
Dawaga, Marwa Elsayh, "Automatic Detection of Eye Blinking Using the Generalized Ising Model" (2016). Electronic Thesis and Dissertation Repository. 4351.