Characterization by Automated System on Infantile Spasms Using Video Analysis and EEG Data

What this Trial Is About

Infantile spasms are sudden, brief epileptic seizures characterized by rapid, repeated body contractions in flexion or extension, sometimes with eye movements. This condition is serious because delayed diagnosis and treatment can lead to cognitive decline. Diagnosing infantile spasms typically requires video-EEG, but this method is costly and not always accessible. To improve early detection, researchers are developing an automated system that analyzes simple smartphone or webcam videos using computer vision and learning models to identify spasms. The study aims to train this automated system by using a database of videos linked with electrophysiological data from pediatric neurophysiology labs. These videos have confirmed spasms based on expert analysis and video-EEG. The goal is for the system to reach over 95% sensitivity and specificity, allowing it to be used by healthcare professionals and families for early seizure detection and monitoring. This approach could facilitate faster referral and treatment, especially where access to specialized diagnostic tools is limited. Participants in this observational study include children up to 3 years old who have undergone video-EEG and have recorded spasms or other epilepsy types. Researchers will analyze existing video and EEG data to teach the system to distinguish spasms from other seizures. The primary outcome is achieving automated detection accuracy above 95% for spasms within two years. The study uses retrospective data from multiple hospitals and plans to publish results in a peer-reviewed journal.

Characterization by Automated System on Infantile Spasmes