Image classification based on ICA-WP feature of EEG signal

Area of Science:

• Neuroscience
• Computer Science
• Signal Processing

Background:

• Electroencephalography (EEG) is crucial for understanding brain activity.
• Classifying visual stimuli from EEG data is challenging due to signal complexity.
• Efficient methods are needed to analyze EEG responses to visual stimuli.

Purpose of the Study:

• To develop and validate a method for classifying EEG recordings based on image stimulation.
• To improve the selection of target images by analyzing EEG data.
• To assess the accuracy and feasibility of the proposed classification technique.

Main Methods:

• Utilizing Independent Component Analysis (ICA) for EEG data dimensionality reduction.
• Employing Wavelet Packet (WP) analysis to capture non-stationary brain activity.
• Combining ICA and WP analysis for feature vector extraction.
• Implementing a Support Vector Machine (SVM) classifier for image classification.

Main Results:

• The combined ICA and WP method effectively extracts feature vectors from EEG recordings.
• The SVM classifier achieved high accuracy in image classification tasks.
• The classification accuracy was minimally affected by variations in classifier parameters.
• A maximum accuracy of 90% was attained, indicating a robust performance.

Conclusions:

• The proposed method is feasible and effective for classifying images based on single EEG stimulations.
• The combination of ICA and WP analysis provides a powerful tool for EEG signal processing.
• This technique offers a promising approach for efficient target image selection in brain-computer interfaces or cognitive studies.