Electroencephalographic compression based on modulated filter banks and wavelet transform.

Area of Science:

• Biomedical Engineering
• Signal Processing

Background:

• Electroencephalographic (EEG) data generates large volumes, necessitating efficient compression for storage, processing, and transmission.
• Lossy compression techniques are crucial for managing the extensive data produced during EEG studies.

Purpose of the Study:

• To evaluate and compare two lossy compression techniques for EEG signals: filter bank decomposition and wavelet packet transformation.
• To identify the optimal compression scheme balancing compression ratio, signal quality, and real-time implementation efficiency.
• To enhance EEG signal compression quality through a proposed quantization stage adapted to signal dynamics.

Main Methods:

• Comparison of filter bank decomposition versus wavelet packet transformation for EEG signal compression.
• Implementation of a novel quantization stage tailored to the dynamic range of individual frequency bands within EEG signals.
• Evaluation of compression performance based on compression ratio, signal fidelity, and computational efficiency for real-time application.

Main Results:

• The filter bank-based compression scheme demonstrated superior performance over transform-based methods.
• The proposed dynamic range-adapted quantization significantly improved the quality of the compressed EEG signals.
• Filter bank compression proved more effective for achieving high compression ratios while maintaining signal integrity.

Conclusions:

• Filter bank decomposition is a more effective approach for lossy compression of EEG signals compared to wavelet packet transformation.
• Adaptive quantization based on signal dynamic range is a key factor in enhancing the quality of compressed EEG data.
• The developed filter bank compression method offers a promising solution for efficient EEG data management in real-time applications.