Presenter Information

Cena Miller, University of Wyoming

Department

Electrical Engineering

First Advisor

Dr. Suresh Muknahallipatna

Description

Brain Computer Interfaces (BCIs) are systems that allow people to control devices using brain signals. One goal of BCI research is to provide an additional communication method for the estimated 7 million Americans who are living with motor impairments, ranging from partial paralysis to limb loss (Center for Disease Control, 2007). Electroencephalogram (EEG) recordings can provide a real-time, low resolution image of the electrical activity on the surface of the brain, using non-invasive techniques. The challenge is in accurately classifying these signals, in order to provide an appropriate command signal to the software interface. The goal of this project was to use EEG data to train a neural network to recognize the patterns in brain activity associated with four types of motor imagery. Then, the network was used to classify data gathered in real-time from the Emotiv Epoc+, a commercially-available EEG headset. Output from the interface was used to control the movement of a shape within a graphical user interface, which also provided feedback about the accuracy of the classification method.

Comments

Poster Presentation

Included in

Education Commons

Share

COinS
 

Electroencephalogram-based Brain-Computer Interface

Brain Computer Interfaces (BCIs) are systems that allow people to control devices using brain signals. One goal of BCI research is to provide an additional communication method for the estimated 7 million Americans who are living with motor impairments, ranging from partial paralysis to limb loss (Center for Disease Control, 2007). Electroencephalogram (EEG) recordings can provide a real-time, low resolution image of the electrical activity on the surface of the brain, using non-invasive techniques. The challenge is in accurately classifying these signals, in order to provide an appropriate command signal to the software interface. The goal of this project was to use EEG data to train a neural network to recognize the patterns in brain activity associated with four types of motor imagery. Then, the network was used to classify data gathered in real-time from the Emotiv Epoc+, a commercially-available EEG headset. Output from the interface was used to control the movement of a shape within a graphical user interface, which also provided feedback about the accuracy of the classification method.