A new approach for EEG feature extraction in P300-based lie detection
Vahid Abootalebi1, Mohammad Hassan Moradi, Mohammad Ali Khalilzadeh
1Electrical Engineering Department, Yazd University, Yazd, Iran. abootalebi@yazduni.ac.ir
Computer Methods and Programs in Biomedicine
|December 2, 2008
Summary
This study enhances the P300-based Guilty Knowledge Test (GKT) for lie detection. It improves accuracy by optimizing feature selection for brain signal analysis, achieving 86% correct detection rates.
Area of Science:
- Neuroscience
- Cognitive Science
- Forensic Psychology
Background:
- Conventional polygraphy faces limitations in accuracy and reliability.
- P300-based Guilty Knowledge Test (GKT) offers a potential alternative using electroencephalography (EEG).
- Previous pattern recognition methods for P300 assessment require further refinement.
Purpose of the Study:
- To extend existing pattern recognition methods for P300 assessment in GKT.
- To enhance the feature set and implement optimal feature selection for improved accuracy.
- To evaluate the performance of the enhanced method in distinguishing guilty and innocent individuals.
Main Methods:
- Subjects underwent a designed GKT paradigm with EEG recording.
- A P300 detection approach utilizing morphological, frequency, and wavelet features was developed.
- A genetic algorithm was employed for optimal feature selection.
- A statistical classifier was used for data classification.
Main Results:
- The enhanced method achieved an 86% correct detection rate for both guilty and innocent subjects.
- This performance surpasses previously reported methods for P300-based GKT.
- The optimized feature set significantly contributed to improved classification accuracy.
Conclusions:
- The extended P300-based GKT with optimized feature selection demonstrates superior performance.
- This approach offers a more accurate and reliable alternative to conventional polygraphy.
- Further research can explore broader applications of this advanced lie detection technique.
