Classification of Electrooculogram signals using Back propagation network – a pilot study

D'Souza, Sandra (2017) Classification of Electrooculogram signals using Back propagation network – a pilot study. In: CISCON 2013, 03/11/2017, MIT, Manipal.

[img] PDF
955.pdf - Published Version
Restricted to Registered users only

Download (838kB) | Request a copy


This paper presents the pilot study for investigating the use of Electrooculography signals (EOG) based classification. The horizontal and vertical EOG signals derived from a group of healthy volunteers (20 male and 20 female ) while performing eye related movements (activities) such as eyes closed, blink, reading and no activity are considered for the proposed study. Features selected from autoregressive models based on Burg, Yule- Walker; linear prediction method and power spectral density estimation using Burg and Yule’s methods are applied and these features are used to classify the age group using a Back-propagation neural network. When features extracted from horizontal EOG are considered, an accuracy of 91.5% with sensitivity of 92.37% and specificity of 90.61% is reported. When classification is performed using features extracted from vertical EOG signals, the accuracy of 84% with sensitivity of 83.84% and specificity of 84.17% is reported. Though the results do not completely support the use of EOG for age classification, by further refining the feature selection process and classification methods, EOG could be used for suitable applications

Item Type: Conference or Workshop Item (Paper)
Uncontrolled Keywords: Electrooculogram, artificial neural network, features, classification, back propagation network.
Subjects: Engineering > MIT Manipal > Instrumentation and Control
Depositing User: MIT Library
Date Deposited: 09 Dec 2017 10:04
Last Modified: 09 Dec 2017 10:04

Actions (login required)

View Item View Item