Award: Phase II STTR Award to Develop an Automated Seizure Detection System

Signal Solutions, LLC Awarded a Phase II NIH STTR Grant to Expand Noninvasive Research Tools for Epilepsy, Seizure and Tremor Detection in Rodent Disease Models.

Signal Solutions has received a two-year Phase II STTR award from the National Institute of Neurological Disorders and Stroke of the National Institutes of Health to develop an automated seizure detection system, in collaboration with Principal Investigator Dr. Sridhar Sunderam, Associate Professor at the F. Joseph Halcomb III, MD, Department of Biomedical Engineering at the University of Kentucky.  The $1.725 million dollar grant, EPIZODE: Noninvasive Seizure Screening in Preclinical Models of Epilepsy, NIH grant number R42NS107148, will fund development of a new method to simplify and streamline seizure detection in animal epilepsy models based on noninvasive piezoelectric sensor technology. This work will expand applications of Signal Solutions’ core technology to other important areas of disease research. The new system will integrate seizure detection with our established sleep/wake tracking system, facilitating the investigation of the sleep-seizure relationship in animal models without the need for invasive surgery.

Epilepsy is marked by spontaneous seizures, and affects over 1% of the population. Seizures also co-occur in other conditions like Alzheimer’s. A third of those with epilepsy do not respond to drugs and 20-30% more experience poor seizure control or side effects. New therapies that prevent or reverse epilepsy are in great demand. The market for epilepsy drugs was $2.9 billion in 2011 and projected to increase by 30% this decade.

Animal epilepsy models are integral to preclinical research. In models of acquired epilepsy, animals typically undergo initial treatment with a convulsant to induce acute status epilepticus—a period of unremitting seizure—followed by a latent period during which the brain rewires itself to generate spontaneously recurring seizures, the hallmark of chronic epilepsy. Therapeutic investigations that target epileptogenesis must first distinguish animals that go on to develop epilepsy after treatment in the early phase of the insult from those that do not. Both the duration of the latent period and the likelihood that an animal will become epileptic are uncertain and animals must be observed for weeks to confirm epilepsy before they are considered ready for invasive EEG monitoring and experimentation. The mortality and long latent period, combined with a variable and uncertain seizure yield after epilepsy is confirmed, make seizure screening a highly resource-intensive process. During the latent period, seizure occurrence is detected and quantitated by visual inspection of video recordings, which can be tedious and error-prone. EEG seizure analysis, the gold standard, requires invasive implantation of electrodes, and is therefore usually done only after confirming onset of epilepsy to accurately detect and quantify seizures.  Genetic models, which may lack a latent period, still must identify the baseline seizure rate against which treatment effects are judged, and would also benefit from a more efficient method of seizure detection.

We established proof-of-concept for detection of epilepsy onset in a chronic epilepsy model in mice and rats. Seizure detection capability of Signal Solutions’ cage-based PiezoSleep^TM system was tested using a pilocarpine model of epilepsy (study funded by NINDS-NIH STTR phase I grant # R41NS107148). The latency period after pilocarpine administration lasts from days to weeks in this model, with a percent of animals typically eventually developing spontaneous recurring seizures. Following pilocarpine administration, piezo signals were screened weekly for seizures using a novelty detector set at a threshold above most behaviors, and seizures were verified by video observation. Animals with good seizure yield were then instrumented for EEG recording and monitored for four more weeks with simultaneous EEG, piezo, and video recording. Seizures detected by EEG were used as the truth set to validate the performance of the piezo detection algorithm. Tonic-clonic seizures have two distinct regimes, with a rhythmic tonic phase followed by a burst-like clonic phase. The piezo signal during a tonic-clonic seizure has distinguishable characteristics during seizure stages: preictal baseline (regular breath signals), rigidity-freezing (tonic phase), rearing/convulsions (clonic phase), post-ictal gasping. Other behaviors show no such structure with largely random fluctuations in signal amplitude and periodicity.  The piezo sensor signature of tonic-clonic seizures was closely correlated with the dynamics of simultaneously measured EEG.  Piezo detected seizures identified 4 out of 5 true seizures (EEG detected seizures), with one out of three piezo-detected events identifying a true event. At this level of sensitivity and specificity, an animal with five seizures in a day requires review of only about one hour of flagged video events to verify true seizures. This method therefore shows great promise, and already a vast improvement over manual video screening, which requires continuous 24-hour review of video recording in the absence of EEG detection methods.

Phase II work is directed toward developing an automated system that captures a range of seizure and tremor behavior. Other animal models of epilepsy show a spectrum of seizure types that reflect the human condition, such as absence, myoclonus, and hypertonic seizures. The SCN8A and GABA(A)γ2R43Q mouse genetic models of childhood epilepsy with absence seizures, and the kainite model of temporal lobe epilepsy, will be used to develop automated piezo-based detection for these seizure types. Essential tremors will be modeled using the Harmaline rat tremor model for tremor detection. The EpiZode software will be incorporated into Signal Solutions’ line of products dedicated to noninvasive methods that simplify research and improve animal welfare.

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

About Signal Solutions

Signal Solutions, LLC specializes in non-invasive solutions for animal research. Our integrated sensor, hardware and software systems are designed to simplify research. The PiezoSleep^TM system is based on EEG-validated piezoelectric sensing of animal movement and behavior, for automated high-throughput scoring of sleep and wake in mice and rats, without the need for surgical implantation for EEG recording.  The complete system includes cage, sensors, hardware and software for data acquisition and analysis and is easily scalable, from a few animals to up to eighty per system, allowing continuous sleep tracking over days or weeks. The PiezoSleep^TM system uniquely enables noninvasive real-time sleep feedback control to improve experimental models, such as sleep disruption, intermittent hypoxia models of sleep apnea, and optogenetic applications. Our system has already been used in areas such as gene and disease model phenotyping, drug testing, and sleep biology for researchers worldwide. Signal Solutions continues to develop new applications and provide custom solutions to simplify and improve methods in other areas of animal research.