Electrooculography-based Eye Movement Classification using Deep Learning Models

Amyotrophic lateral sclerosis (ALS), also known as motor neuron disease (MND), is a specific disease that causes the death of neurons controlling voluntary muscles. Most ALS patients eventually lose the ability to walk, use their hands, speak, swallow, and breathe. In this paper, we use the electrooculogram (EOG) signals captured using four sensors placed on the controlling muscles of the eye movement in horizontal and vertical directions to classify four different eye movements. The classifier output is used to control a wheelchair, or any other device developed to help ALS patients in performing their daily needs. Contrary to the classical classification techniques where features are extracted first from the EOG signals, then used with a trained classifier, in this paper the EOG signals are fed directly into two deep neural networks using, respectively, the long-short term memory (LSTM) and the convolutional neural network (CNN). The results show an accuracy of 88.33% for the LSTM network and 90.3% for the CNN network in eye movement classification.