Overview
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. Their 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).
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 SVM (Support Vector Machines). An overview of the final BCI system protocol is shown below.
- 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
- 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.