LAMPSY - A Novel Epilepsy Video Monitoring and Seizure Detection Device

Vicente Mayer Mendonça de Lima Garção

Key information

Authors:

Vicente Mayer Mendonça de Lima Garção (Vicente Mayer Mendonça de Lima Garção)

Supervisors:

Hugo Humberto Plácido da Silva (Hugo Humberto Plácido da Silva); Ana Luísa Nobre Fred (Ana Luísa Nobre Fred)

Published in

12/07/2021

Abstract

Epilepsy is a neurological condition characterized by the occurrence of unprovoked seizures. It affects 50 million people worldwide, 30% of which have pharmacoresistant epilepsy, leading to frequent seizures that may considerably worsen their quality of life and potentially lead to premature death. Seizure detection and registration could be essential in facilitating the diagnosis and management of this condition and improving diagnostic accuracy, thereby enabling better medication adjustment. It can also reduce stress levels in patients and caregivers and ensure that timely help can be delivered when dangerous seizures happen. The main objective of this work was to create an accurate, affordable, unobtrusive, and accessible video-based seizure detection device that preserved patient privacy by design. This device was named "Lampsy", and a patent request for it was filed. A prototype was built using a Raspberry Pi single-board computer integrated within a light fixture, a small camera, and a display, and placed in Hospital de Santa Maria for video acquisition during video-EEG sessions. The video data was transmitted to a secure server via SFTP, preserving patient privacy. A seizure detection algorithm based on optical flow, PCA and ICA-based seizure movement isolation, novel seizure metrics, and machine learning classification was developed. High detection accuracy was achieved, with an AUC of 96.5%, sensitivity, and specificity of 94.8% at the equal error rate, and an average latency of 52.9 seconds. Seizures were segmented with an error of +/- 8.85 seconds when compared to annotations by health professionals.