Skip to content

Overview

Faheem Ershad edited this page Jun 29, 2018 · 12 revisions

Brain computer interface (BCI) using dry electrode EEG

This is a 2017/2018 Capstone Project from the Department of Biomedical Engineering at the University of Houston. The members of this team were Faheem Ershad, Ethan Hart, Zhihe (Harrison) Zhao, and Kaitlyn Robinson. Our advisors were Dr. Sridhar Madala, CEO of Indus Instruments (Webster, TX) and Dr. Joe Francis, Associate Professor in the Department of Biomedical Engineering at the University of Houston (Houston, TX). Mohammad Badri, Thinh Nguyen, and Tom Potter (graduate students at the time of writing) provided unwavering support to the group throughout the duration of the project and we thank them and our advisors for their guidance.

The objective of this project was to create a working brain-computer interface (BCI) that combines a dry-electrode EEG headset and online/offline classifier algorithms to control a computer cursor. The classifier algorithms will extract relevant features from the EEG data using PSD (Power Spectral Density) and CSP (Common Spatial Patterns) and then classify with SVMs (Support Vector Machines). An overview of the final process is shown below.

Hardware

  • 3D printed headset connected to a Cyton Biosensing board (32 bit, 8 channel) with 5 reusable, spiky, dry electrodes over Fc1, Fc2, C3, C4, and Cz from the 10-20 system for EEG
  • Cyton board connects to PC via USB RFDuino radio dongle

Software

  • OpenViBE for data acquisition and offline analysis
  • MATLAB for feature extraction and classification
  • MATLAB for robot cursor control and corresponding GUI

Two versions of classifiers were made for online prediction of left/right/non- hand movement. The first was a left/right classifier. The second had a not moving/moving classifier followed by a left/right classifier. The diagram for this is shown here.

Check out the Branches Wiki page next!

Clone this wiki locally