Abstract:
The phenomenon of the increase in alpha EEG activity associated with eye closure has been shown to be successful for implementing environmental control
for disabled persons. Studies in this thesis investigate strategies which improve the reliability, robustness, and ease of use of alpha EEG control systems. Primarily, research covers the effectiveness of alpha EEG detection algorithms (with regard to detection time and susceptibility to artifact) and the construction and use of EEG sensing electrodes.
Many new techniques for the detection of the increase of alpha EEG associated
with eye closure are researched, developed, implemented and evaluated.
All detection techniques are compared to a conventional method using a novel performance parameterisation criterion. In conjunction with the application of the same EEG data sets to all techniques, the use of the performance criteria enables a fair and quantitative comparison to be made between alpha detection methodologies. Detection techniques employed include
enhanced versions of conventional methods, localisation of apparent alpha sources in the brain, and preprocessing methods (such as spatial filtering,
adaptive filtering and independent component analysis). The best performance of alpha EEG detection was given by the source power alpha localisation technique, which showed statistically significant and practically important improvements in performance over conventional techniques. Additionally, this localisation technique is convenient and fast to implement.
In situations in which electrodes are intended for unsupervised use with environmental control systems, the evaluation of alternative electrode types to the conventional wet electrodes is required, as the use of wet electrodes has several drawbacks. The performance of wet, dry and insulating electrodes is compared in this research. One aspect of the quantitative comparison of electrodes types is the measurement of contact impedance. To enable the fast and accurate measurement of impedance spectra, a new impedance spectroscopy system was developed as part of this thesis. In addition to comparison of impedance criteria, electrodes were evaluated in the presence movement-based, and electric field induced, artifacts. The electrode comparisons were carried out in a direct and quantitative manner in a controlled test environment for the first time. Results indicate that, in contrast to earlier reports, both dry and insulating electrode perform well with respect to artifact and offer a viable alternative to wet electrodes for long-term monitoring of biosignals from the surface of the skin. More improvements are required before such electrodes are suitable for EEG usage.
