Neural Network based Eye Position Estimation from Infrared Video during Sleeping

Abstract

To analyze relationship between brain function and eye movement during sleeping, a simultaneous measurement system using functional magnetic resonance imaging (fMRI) and infrared video oculography has developed. However, it takes a lot of time, and requires considerable efforts to manually detect eye movements from the infrared video. In this paper, we propose an automated method for estimating eye position using neural network. The effectiveness of the proposed method is demonstrated by applying our method to a subject who is sleeping with REMs (rapid eye movements).