Development of signal processing algorithm for familiarity detection from brain signal

• Main Author: Tan, Ernest Zheng Hui
• Other Authors: Vinod Achutavarrier Prasad
• Format: Final Year Project
• Language: English
• Published: 2015
• Subjects: DRNTU::Engineering::Computer science and engineering::Computer systems organization::Computer system implementation; DRNTU::Engineering::Computer science and engineering::Computing methodologies::Pattern recognition; DRNTU::Engineering::Computer science and engineering::Computer applications::Life and medical sciences
• Online Access: http://hdl.handle.net/10356/62618
• Institution: Nanyang Technological University

Electroencephalography is a well-documented phenomena with extensive research done on its properties such as memory recollection. Despite studies done on EEG memory properties, the bulk of these studies did not examine the problem from Time-Frequency (TF) domain perspective. Therefore this study aims to examine the feasibility of familiarity detection in TF by developing a signal processing algorithm for feature extraction from EEG oscillations. Experiments in this study were conducted using an Emotiv EPOC, a .NET application and MATLAB. Results suggested manifestation of familiarity effect in the EEG signals. The extracted features were also found to have varying success in familiarity detection with the Peaks and Complexity combination having 67.15% mean classification accuracy using SVM. These findings demonstrates the feasibility of familiarity detection in TF however having rest periods in the experiment protocol is recommended to reduce the possibility of residual neural interference. Improvements in extraction techniques should also be looked into.