Video-based automated detection of generalized tonic-clonic seizures using deep learning
Background
Detecting epileptic seizures in a timely manner is crucially important as early detection can prevent complications such as secondary injuries, falls, and neurological damage. Early detection also allows for timely treatment and care that can stop or limit a seizure’s extent. In patients with generalized tonic-clonic seizures (GTCS), timely detection of seizures is especially important as sudden death in epilepsy (SUDEP) is more common in patients with GTCS.
Video-EEG seizure monitoring is considered the gold-standard in detecting epileptic seizures. It is routinely performed in specialized epileptic monitoring units due to its essential role in the diagnosis and treatment of epilepsy. However, this monitoring and evaluation system is very time and labor-consuming. It is also often limited, due to resource constraints, in both the hospital and home settings. Automated video-based seizure detection methods would allow a better monitoring and optimization of seizures and treatments. It would decrease complications, such as the risk of SUDEP. It would allow for a major decrease in the time and labor involved in screening and evaluating long-term video data in specialized epileptic monitoring units.
Technology Overview
This invention discloses a self-contained, automated video detection system to detect GTCS and alarm caregivers of a seizure occurring in real-time. It is composed of a device to continuously record video, a processing unit to continuously analyze the video data, and an output unit that provides an alarm when an epileptic video is detected. In its entirety, this invention provides an apparatus that continuously, and in real-time, monitors an epileptic patient in the home or hospital setting and provides means for alerting caregivers or clinicians of a dangerous status.
Benefits
- Can be used when a patient is sleeping or awake
- Can be used in a home or hospital setting
- Decreases complications, such as the risk of SUDEP
- Allows for a major decrease in the time and labor involved in screening and evaluating long-term video data in specialized epileptic monitoring units
- Real-time, continuous monitoring and analysis
- Lower cost than a non-continuous or non-real-time seizure detection and alarm system
Applications
- The monitoring of and early detection of epileptic episodes
Publications:
Yang Y, Sarkis RA, Atrache RE, Loddenkemper T, Meisel C. Video-Based Detection of Generalized Tonic-Clonic Seizures Using Deep Learning. IEEE J Biomed Health Inform. 2021 Aug;25(8):2997-3008. doi: 10.1109/JBHI.2021.3049649. Epub 2021 Aug 5. PMID: 33406048.
