Lossless EEG Compression Using the DCT and the Huffman Coding

Abstract

The telemedicine and ambulatory monitoring development motivates many researchers to focus on the lossless electroencephalogram (EEG) compression in recent years. Nevertheless, most of these studies could not present the potential of compression techniques such as discrete cosine transform (DCT), and Huffman coding due to lack of attention to the signal and technique characteristics. In this work, we developed a lossless hybrid EEG compression method based on the characteristic of DCT frequency spectrum and the Huffman coding. In our method, we calculate the DCT coefficients below <em>40</em> Hz (dominant components) of the EEG segments. Then, we code these quantized DCT coefficients using a Huffman coder in the transmitter site. In the receiver site, we add a zero set for the DCT coefficients above <em>40</em> Hz, and then reconstruct the EEG segments using the inverse DCT. We applied our method for the five sets (denoted A-E) of the Bone University database. The results indicate that our algorithm can improve the average compression ratio of these sets up to <em>1.78, 1.94, 2.66, 3.35</em>, and <em>1.78</em> times of the best results in the literature, respectively. Therefore, our hybrid method could compress the single-channel signal satisfactory enough.