An evaluation of the effects of wavelet coefficient quantisation in transform based EEG compression

Area of Science:

• Biomedical Engineering
• Signal Processing
• Medical Informatics

Background:

• Telemedical and ambulatory applications require efficient electroencephalogram (EEG) signal compression.
• Lossy compression techniques can significantly reduce data size but risk information loss.
• Quantization and thresholding are common methods for data reduction in signal processing.

Purpose of the Study:

• To compare the effectiveness of adjusting quantization levels versus standard thresholding within the SPIHT algorithm for EEG signal compression.
• To evaluate the impact of these compression strategies on signal fidelity and data reduction efficiency.

Main Methods:

• The study employed the Set Partitioning In Ideal Trees (SPIHT) algorithm, a wavelet-based compression method.
• Compression was achieved by modifying the quantization levels of Discrete Wavelet Transform (DWT) coefficients and comparing it to SPIHT's standard thresholding.
• The performance was evaluated on electroencephalogram (EEG) signals.

Main Results:

• Increasing quantization levels in SPIHT, combined with its entropy encoding, yielded significantly better results than standard SPIHT alone.
• This combined approach demonstrated improved data compression efficiency for EEG signals.
• The study identified an optimal balance between data reduction and signal fidelity through controlled quantization.

Conclusions:

• The optimized SPIHT algorithm, incorporating enhanced quantization, provides a superior method for compressing EEG signals.
• This technique is highly suitable for telemedical and ambulatory EEG applications demanding efficient data transmission.
• Further research can explore adaptive quantization strategies for even finer control over EEG compression.