Related Experiment Video
Updated: Jul 30, 2026
09:43
Video-oculography in Mice
Published on: July 19, 2012
23.8K
Automatic Video-Oculography System for Detection of Minimal Hepatic Encephalopathy Using Machine Learning Tools
Alberto Calvo Córdoba1, Cecilia E García Cena2, Carmina Montoliu3,4
1Escuela Técnica Superior de Ingenieros Industriales, Center for Automation and Robotics, UPM-CSIC, Universidad Politécnica de Madrid, José Gutiérrez Abascal St., 2, 28006 Madrid, Spain.
Sensors (Basel, Switzerland)
|October 14, 2023
Summary
An automatic eye-tracking system using machine learning can detect minimal hepatic encephalopathy (MHE) in cirrhotic patients. This novel video-oculography method is faster and as accurate as traditional tests for diagnosing this liver disease complication.
Area of Science:
- Neurology
- Biomedical Engineering
- Machine Learning
Background:
- Minimal hepatic encephalopathy (MHE) is a common neurological complication in liver disease patients.
- Current diagnostic methods like the Psychometric Hepatic Encephalopathy Score (PHES) are time-consuming (25-40 min) and require expert interpretation.
- There is a need for efficient and objective diagnostic tools for MHE.
Purpose of the Study:
- To present an automatic gaze-tracker system for MHE detection.
- To utilize machine learning and video-oculography for analyzing eye movements.
- To offer a faster, more accessible diagnostic alternative for MHE.
Main Methods:
- Developed automatic feature-extraction software and a machine learning algorithm using video-oculography.
- Collected eye movement data from 47 cirrhotic patients, classified using the PHES as the gold standard.
- Analyzed 150 eye movement features, selecting the most relevant (p ≤ 0.05) for algorithm training.
Main Results:
- The automatic video oculography system is a simple test, taking 7-10 min per patient.
- The system achieved 93% sensitivity and 93% specificity in MHE detection.
- This method offers a significant improvement in efficiency compared to the PHES battery.
Conclusions:
- Automatic video oculography with machine learning provides a rapid and accurate method for MHE detection.
- This technology can assist clinicians in diagnosing MHE more effectively.
- The system is suitable for longitudinal analysis, unlike the PHES.
