Analog implementation of seizure detection algorithm

• Published in: Proceedings of the First Joint BMES/EMBS Conference : serving humanity advancing technology, Oct. 13-16, 99, Atlanta, GA, USA Vol. 2; pp. 860 vol.2 - 860
• Main Authors: Nikitin, A.V., Doynov, P., Davidchack, R.L., Frei, M.G., Osorio, I.
• Format: Conference Proceeding Journal Article
• Language: English
• Published: IEEE 1999
• Subjects: Algorithm design and analysis; Algorithms; Detection algorithms; Digital filters; Epilepsy; Finite impulse response filter; Monitoring; Robustness; Signal detection; Signal filtering and prediction; Signal processing; Sorting; Statistics
• ISBN: 0780356748; 9780780356740; 9780780356757; 0780356756
• ISSN: 1094-687X; 0589-1019; 1558-4615
• DOI: 10.1109/IEMBS.1999.804015

An analog device for automated real-time detection and short-term prediction of epileptic seizures is described. In its basic form, the device utilizes a non-sorting analog percentile filtering to implement in analog form the FHS numerical seizure detection algorithm. The extensions and enhancements of the basic version of the device include (1) adjustment (individual adaptation) of the front-end filter for pre-processing the ECoG signal, or combining several filters in a 'filter bank' to increase the reliability of detection, (2) adjustment of the parameters of the analog percentile filter to emphasize desirable signal features, and (3) utilization of a multi-channel percentile filter module with adjustable parameters. The described device can easily satisfy size, weight, and power criteria applicable to portable and/or implantable devices and combines the robustness and reliability of a generic algorithm with an easy adjustment (adaptation) to individual patients. Development of such a device will translate into better care for persons with epilepsy in a cost effective manner by minimising risk of injury or sudden death and improving quality of life by freeing them to pursue normal activities, including work.