Detection of focal epileptiform activity in the EEG: an SVD and dipole model approach
Type of content
UC permalink
Publisher's DOI/URI
Thesis discipline
Degree name
Publisher
Journal Title
Journal ISSN
Volume Title
Language
Date
Authors
Abstract
An algorithm has been developed for detection of epileptiform activity in the EEG. The EEG is divided into overlapping epochs, which undergo two steps. The first is singular value decomposition (SVD) which identifles the number of uncorrelated active sources sources in an epoch.In the second step, EEG dipole source analysis, using a single dipole model, is applied to the EEG. This yields dipole parameters and a relative residual energy (RRE). The detection algorithm triggers an EEG epoch when SVD indicates a dominant source and the RRE is low. The algorithm is applied to simulated EEG generated by two sources which are synchronously and asychronously active. For the synchronous case the critical measure is the RRE whereas for the asynchronous case both the SVD and RRE are critical. The algorithm has also been applied to real EEG containing two spikes and an eye-blink artifact. The SVD indicated a dominant active source and the RRE was low for all three events. These preliminary results demonstrate the potential of the method for detection of spikes and seizures with a focal origin.