A Master of Science thesis in Electrical Engineering by Salma Khaled Mohamad Zeid entitled, "Measurement of Vigilance Using EEG Source Localization," submitted in November 2017. Thesis advisor is Dr. Hasan Al-Nashash and thesis co-advisor is Dr. Hasan Mir. Soft and hard copy available.
Vigilance, or sustained attention, is crucial for jobs where attentiveness for prolonged times is required. These jobs include air traffic control, luggage inspection, and surveillance jobs. Vigilance decrement can cause catastrophic consequences. Therefore, vigilance level assessment is a widely-researched topic. Several methods have been used to assess vigilance levels such as eye tracking techniques which include monitoring saccadic eye movements and pupil size variation. Other methods used are heart rate variability, and physiological data such as electrocardiogram (ECG), electro-oculogram (EoG) and electroencephalogram (EEG). EEG data has been found to have strong correlations with human's vigilance level. This thesis report presents a novel method for the assessment of vigilance decrement using EEG data that embarks upon the brain's temporal behavior. An experiment based on a 20 to 30-minute Psychomotor Vigilance Task (PVT), that simulates real applications where vigilance decrement is observed, was carried out on 33 subjects and their EEG recordings and reaction times were collected. In the PVT task, subjects were required to respond to target events while refraining from non-target events. Vigilance reinforcement by challenge integration was tested where 22 out of the 33 subjects had an additional task where they had to respond to noisy target events. The spectral power density characteristics namely the delta, theta, alpha and beta waves of the EEG data are compared for low and high vigilance states. Furthermore, EEG source localization is utilized to monitor source dynamics of the brain in transition from vigilance states. Results from both methods are analyzed using Student's t-test with the significance threshold set at 0.1. Power spectral density analysis showed that power in AF8 electrode in delta, theta and alpha bands increased with vigilance decrement with p-values .023, .079 and .020 respectively. The source localization approach showed an increase in prefrontal source distribution with vigilance decrement with p-value of .015. The joint probability function of the prefrontal delta, theta and alpha bands as well as the source dynamics of prefrontal activity showed promise in constructing a vigilance assessment model to identify vigilance state from labeled data by yielding an 84.85% accurate detection.