Seizure recognition and analysis.

The processing of EEGs with respect to epileptic seizures falls in two main categories: automatic recognition of seizures during long-term monitoring and analysis of seizures to extract information not available from visual inspection. Automatic seizure recognition is a complex problem because the EEG during seizures is not well defined morphologically; imperfect automatic recognition is nevertheless possible and it can simplify the task of monitoring and increase its diagnostic yield. The widespread use of monitoring techniques has made the recording of seizures relatively frequent. It has therefore become worthwhile to develop procedures for analyzing seizure patterns in detail, particularly the propagation of seizure activity between different brain structures. The computation of time differences of a few milliseconds between two channels is possible and has been shown to be meaningful. Such an analysis allows to follow the structures involved and the different stages of evolution of a seizure, the structures likely to be 'driving' seizure activity and the possible routes of propagation. These methods are reviewed and their use is illustrated.