Independent component analysis in the distinction between focal and multifocal spikes.

Abstract

Independent component analysis (ICA) is a novel method that extracts independent sources in recorded signals. One of its capabilities is to separate epileptiform activity of different origins. The goal of this study was to demonstrate that ICA is useful for differentiating focal and multifocal epilepsies. Using ICA, the authors analyzed 160 samples of patients with unifocal (temporal: 50 samples from 5 patients; frontal: 50 samples from 5 patients) or multifocal epilepsy (bitemporal: 30 samples from 3 patients; multifocal extratemporal: 30 samples from 5 patients). Each sample included at least two spikes. ICA was applied using the JADE algorithm implemented in a Matlab platform. The components were identified visually. The EEG and the isopotential map of the suspected components were reconstructed to demonstrate the nature and location of each spike. In multifocal epilepsies, the spikes were separated in distinct components in all cases. In unifocal temporal epilepsies, ICA extracted all the spikes from the same location into a single component. In the patients with unifocal frontal epilepsy, one component included all the spikes in 80% of the samples; in some of these cases, other components were responsible for small parts of the spikes, but with a close topography to the main component. The waves were separated in different components both in the unifocal and in the multifocal samples. Therefore, in epilepsies with independent foci, ICA separates the spikes of different origins into distinct components. In unifocal epilepsies, ICA tends to locate the spikes from the same area into the same component, especially in temporal epilepsies. The authors conclude that ICA might be a useful tool to distinguish between unifocal and multifocal epilepsies.