Electronic Theses and Dissertations

Identifier

1403

Date

2015

Date of Award

5-27-2015

Document Type

Thesis

Degree Name

Master of Science

Major

Electrical and Computer Engr

Concentration

Computer Engineering

Committee Chair

Bashir I Morshed

Committee Member

Jacobs L Eddie

Committee Member

Amy L De Jongh Curry

Abstract

Electroencephalography (EEG) is a promising technique to record brain activities in natural settings. EEG signal usually gets contaminated by Ocular Artifacts (OA), removal of which is critical for the feature extraction and classification. With the increasing interest in wearable technologies, single channel EEG systems are becoming more prevalent that often require real-time signal processing for immediate feedback. In this context, a new hybrid algorithm to detect OA and subsequently remove OA from single channel streaming EEG data is proposed here. The algorithm first detects the OA zones using Algebraic approach, and then removes artifact from the detected OA zones using Discrete Wavelet Transform (DWT) decomposition method. De-noising technique is applied only to the OA zone that minimizes interference to neural information outside of OA zone. The microcontroller hardware implemented hybrid OA removal algorithm demonstrated real-time execution with sufficient accuracy in both OA detection and removal. The performance evaluation was carried out qualitatively and quantitatively for 0.5 sec epoch in overlapping manner using time-frequency analysis, mean square coherence, Correlation Coefficient (CC) and Mutual Information statistics. Matlab implementation resulted in average CC of 0.3242 and average MI of 1.0042, while microcontroller implementation resulted in average CC of 0.4033 for all blinks. Successful implementation of OA removal from single channel real-time EEG data using the proposed algorithm shows promise for real-time feedabck applications of wearable EEG devices.

Comments

Data is provided by the student.

Library Comment

dissertation or thesis originally submitted to the local University of Memphis Electronic Theses & dissertation (ETD) Repository.

Share

COinS