Thomas E. Peters

Performance reassessment of a real-time seizure-detection algorithm on long ECoG series

Summary:  Purpose: Automated seizure detection and blockage requires highly sensitive and specific algorithms. This study reassessed the performance of an algorithm by using a more extensive database than that of a previous study and its suitability for safety/efficacy closed‐loop studies to block seizures in humans.

Network system for automated seizure detection and contingent delivery of therapy

Summary The authors describe an integrated bedside system for real-time seizure detection and automated delivery of electrical stimulation directly to the brain of subjects undergoing invasive epilepsy surgery evaluation. These stimulations were triggered by specific detections following a prespecified pattern. The system uses a commercially available EEG unit, two personal computers, two Grass S-12 stimulators, and other custom-built units to enable interfacing between these components. To date, more than 9,500 hours of electrocorticographic data have been acquired, displayed, and analyzed, and more than 900 closed-loop stimulations for seizure blockage have been delivered safely and reliably to eight subjects with intractable epilepsy. This system generates on-line reports containing information about seizures that provide the epileptologist with timely, valuable data while allowing adaptation of the algorithm detection parameters to improve its performance if necessary. Additionally, it can control the output of any therapeutic device and administrate automatically cognitive tests or radioactive tracers for neuroimaging purposes. This network system, which can be replicated at a relatively low cost by others, is proof of concept for a portable or implantable device that could serve identical functions. Widespread availability of this type of system will advance the fields of clinical and basic epilepsy rapidly and considerably.