Abstracts

Epilespy research in animals is hampered by the inability to efficiently document electrophysiological and behavioral manifestations of epilepsy. Our objective was to develop a practical, high volume, automated laboratory for unattended monitoring of mulitple animals.
The RMU (rodent monitoring unit) is designed for continuous monitoring of cortical, depth, or microelectrodes from 21 animals simultaneously. All cages are compliant with the ILAR Guide, so that the animals are actually housed in the monitoring cages for chronic experiments. Systems acquire up to10 EEG / ECoG channels per [ldquo]bed[rdquo] via input connections from animal to amplifier using commutator swivels, which allow the animals to freely move about their cages. Full motion (30 /sec) synchronized digital video (MPEG) is acquired. Review stations display and process the EEG and video data from any of the animals over a 100 Mbit Ethernet network. The systems are designed with 100 hours of raw EEG and video storage, plus additional room for saved segments. Individually addressable stimulators for 10 of the beds are computer-controlled, and on-line seizure detection permits closed-loop control of stimulators.
Intensive evaluation of data quality was carried out on 10 beds over several monhts by scoring each day of recording as evaluable or not. A day was considered non-evaluable if the artefacts involved both hippocampal electrodes at the same time, and the artifacts were present through more than 25% of the time.
Prepatory to experiments on the efficacy of closed loop stimulation for seizure control, the invasive seizure detection program was pilot tested on a rat model of temporal lobe epilepsy displaying spontaneous seizures. During a 24 hour period in one rat, automatic seizure detection results were compared to seizures that had already been visually identified and manually marked by a trained technician.
The systems have been used daily and updated over the last 4 years to study a variety of animal models of epilepsy. Since its inception, occupancy of the RMU beds has averaged 79%. Investigators acquire and review data simultaneously with no data interruption or effect on performance.
Out of a total of 383 rat monitoring days (days times number of rats on each day) in the data integrity study, 18 (4.7%) were judged non-evaluable due to artifacts (primarily due to disconnection or chewed lead wires).
In the 24 hour seizure-detector dataset, there were 32 spontaneous seizures. When applied to retrospective ECoG data in an off-line fashion, the seizure detector runs at 15 times real time. Using the default settings, the program detected 22 of the seizures (69%) with zero false positives. Optimizing the sensitivity improved the true positive detection to 89% with five false negatives.
High volume, computer-assisted video-EEG facilitates chronic animal research on neurophysiological mechanisms and treatment of epilepsy.