Classification of brain electrophysiological changes in response to colour stimuli

Area of Science:

• Neuroscience
• Brain-Computer Interfaces
• Signal Processing

Background:

• Classifying visual stimuli from brain activity is crucial for understanding perception.
• Previous research has primarily focused on event-related potentials (ERPs), leaving ongoing brain activity classification under-explored.
• Distinguishing colours based on electroencephalography (EEG) signals has not been previously reported.

Purpose of the Study:

• To classify ongoing EEG brain responses to visual stimuli of blue, green, and red colours.
• To investigate the efficacy of using ongoing EEG signals versus ERP metrics for colour classification.
• To identify brain regions involved in colour categorization using EEG data.

Main Methods:

• Feature extraction using Fourier transform to obtain EEG band power values (delta, theta, alpha, beta, gamma).
• Utilized both scalp EEG measurements and source-space features derived from inverse solutions.
• Applied one-way ANOVA for feature selection and implemented four classifiers, including KNN and Ensemble-KNN.

Main Results:

• The classification of RGB colours was achieved with 93% accuracy using one-second EEG data.
• Ensemble-KNN and KNN classifiers demonstrated the highest accuracy with both scalp and source-space features.
• Analysis indicated that colour perception involves brain regions beyond the primary visual cortex, particularly temporal and temporoparietal areas.

Conclusions:

• Ongoing EEG signal analysis is a viable method for classifying visual stimuli like colours.
• The findings highlight the involvement of distributed brain networks in colour perception.
• High accuracy in colour classification using EEG opens avenues for advanced brain-computer interfaces.