Abstracts

Rationale: A growing number of research groups worldwide works on the application of advanced analysis methods for EEG data. Long-term EEG data are of particular interest e.g. for the development and evaluation of spike and seizure detection algorithms as well as for time series analyses for seizure prediction. Available EEG data are scarce. We here report progress in the development of a European EEG database containing abundant metadata as part of the European Project EPILEPSIAE (www.epilepsiae.eu). This work has been approved by the ethics committies of the participating centers and is supported by the European Union (grant 211713). Methods: A relational database scheme was developed for annotated long-term EEG data, derived features, MRI data, and clinical metadata including seizure time points, patterns and spread of ictal activity, and typical spike topographies. Data from three European epilepsy centers (Coimbra, Freiburg and Paris) are being integrated into the database. A web client was programmed for queries on predefined datasets and for accessing data. Results: As of June 2010, more than 150 patients with long-term continuos EEG datasets of at least 4 days duration have been included in the database. These data include more than 1000 clinical ictal events as well as interictal periods of interictal EEG. Patients both with surface and intracranial EEG have been integrated. Presently, EEG data at higher sampling frequencies of 1000-2500 kHz are additionally introduced. Conclusions: There is considerable progress in this database which sets a new standard for EEG time series analyses and allows for new applications, e.g. subgroup analyses of algorithms for particular EEG syndromes, and intraindividual training and testing of algorithmic performance. It is considered to open access to this European database to research groups from 2012 on.