Skip to content

High School SSVEP-BCI Research Project to improve classification accuracy of captured EEG signals

License

Notifications You must be signed in to change notification settings

RandyDeng/SSVEP-BCI-Research

Repository files navigation

SSVEP-BCI Research

As part of a professional research program at Ridgefield High School, I researched and conducted experiments in the field of brain-computer interfaces. Under the mentorship of Dr. Dean Krusienski and his PhD students at ASPEN Lab in Old Dominion University (now moved to Virginia Commonwealth University), I measured human steady-state visually evoked potentials (SSVEP) using an electroencephalogram (EEG). The SSVEPs were generated by providing flashing checkerboard images to the subjects and could potentially be used to assist disabled patients interact with their environment without moving.

Abstract:

A major challenge since the invention of the steady-state visually evoked potential (SSVEP)-based Brain-Computer Interface (BCI) has been improving accuracy and signal recognition. Although SSVEPs have exhibited high accuracy rates in subjects with minimal BCI exposure, to be reliable for everyday use, BCIs must achieve high, if not 100% accuracy. In this study, we examine the effect of altering the size of the checkerboard pattern on the SSVEP signal at 6Hz and 10Hz. The size of the pattern was evaluated on a continuum from a large pattern, which is equivalent to a solid flashing stimulus, to a bounded single pixel checkerboard (256x256 pixels) with the same boundary. The boundary was a 256x256 pixel square. The number of checkerboard tiles quadrupled with each increase (the number of checkerboard tiles in each side was doubled), resulting in the following checkerboard sizes: 1x1, 2x2, 4x4, 8x8, 16x16, 32x32, 64x64, 128x128, and 256x256 (pixel size). A Fast Fourier Transform was done to graphically display the power spectral density (PSD) of the SSVEP signals and a paired t-test was done between the increasing checkerboard and solid stimuli with their respective frequencies to see if there was any significant power difference. Results indicate that 2x2 and 4x4 stimuli generally create the most distinct SSVEP signal, which becomes less noticeable as the checkerboard stimuli size became smaller.

About

High School SSVEP-BCI Research Project to improve classification accuracy of captured EEG signals

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published