Selection of Optimal wavelet for lossless EEG compression for real-time applications

This paper discusses a wavelet based, lossless compression scheme for Electroencephalogram(EEG) signals. For any wavelet based compression scheme, selection of wavelet base is very important task for obtaining effective signal compression. Here, emphasis is given for selection of wavelet basis and number of decomposition levels. The proposed EEG compression algorithm comprises of a preprocessing step involving computation of the difference signal of EEG time series, followed by integer lifting wavelet transform and set partitioning in hierarchical trees (SPIHT) as a wavelet encoder. The optimal wavelet base is chosen based on the compression ratio (CR) as well as time taken for encoding the wavelet coefficients by SPIHT coder. The algorithm is tested with normal and abnormal multi-channel electroencephalogram data of 8 minutes duration. The optimal wavelet base and decomposition level is presented based on the experimental results.