we propose to develop a completely functioning prototype system for automated and high-throughput seizure detection and classification in animals, mainly in mice or rats or guinea pigs, based on semiology (video record). The purposes of the project are to first develop technologies and tools that can automatically detect seizures through visual observation; then further classify those seizures according to a classification scheme when possible. The video detection and classification approach will complement EEG analysis approach to achieve accurate and precise results. These detection and classification will be conducted in high throughput mode to meet the increasing demand. This technology will open a new avenue to detecting and analyzing seizures besides traditional ictal epileptiform (EEG) recording analysis, and can greatly improve the results of detection and classification of seizures in rodents, thus providing an extremely powerful tool in epilepsy research. This technology will also improve the quality and alleviate the burden of observing rodents' seizures and achieve more objective and consistent analysis. The technology will also contribute to homeland security for chemical defense. For this purpose, this project will include studies in the areas of behavioral research, epilepsy and seizure observation and analysis, computer algorithm design, as well as complete system integration. We will make use of the technologies for digital video processing and analysis that we have developed with past DARPA projects and our current NIMH and NIDA projects.
