The feasibility of implementing a real-time system for removing eyeblink artifacts from electroencephalogram (EEG) recordings utilizing a hardware/software coprocessing platform was investigated. A software based wavelet and independent component analysis (ICA) eyeblink detection and removal process was extended to enable variation in its processing parameters. Exploiting the efficiency of hardware and the reconfigurability of software, it was ported to a field programmable gate array (FPGA) development platform which was found to be capable of implementing the revised algorithm, although not in real-time. The implemented hardware and software solution was applied to a collection of both simulated and clinically acquired EEG data with known artifact and waveform characteristics to assess its speed and accuracy. Configured for optimal accuracy in terms of minimal false positives and negatives as well as maintaining the integrity of the underlying EEG, especially when encountering EEG waveform patterns with an appearance similar to eyeblink artifacts, the system was capable of processing a 10 second EEG epoch in an average of 123 seconds. Configured for efficiency, but with diminished accuracy, the system required an average of 34 seconds. Varying the ICA contrast function showed that the gaussian nonlinearity provided the best combination of reliability and accuracy, albeit with a long execution time. The cubic nonlinearity was fast, but unreliable, while the hyperbolic tangent contrast function frequently diverged. It is believed that the utilization of programmable logic with increased logic capacity and processing speed may enable this approach to achieve the objective of real-time operation.
Library of Congress Subject Headings
Electroencephalography--Data processing; Eye--Movements--Data processing; Wavelets (Mathematics); Field programmable gate arrays
Department, Program, or Center
Electrical Engineering (KGCOE)
Chaudhuri, Matthew Alan, "Optimization of a hardware/software coprocessing platform for EEG eyeblink detection and removal" (2008). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus