Optimal linear prediction for the lossless compression of volume data

Summary form only given. Data in volume form consumes an extraordinary amount of storage space. For efficient storage and transmission of such data, compression algorithms are imperative. However, most volumetric data sets are used in biomedicine and other scientific applications where lossy compression is unacceptable. We present a lossless data compression algorithm which uses optimal linear prediction to exploit correlations in all three dimensions. Our algorithm is a combination of differential pulse-code modulation (DPCM) and Huffman coding and results in compression of around 50% for a set of volume data files. The compression algorithm was run with each of the different predictors on a set of volumes consisting of MRI images, CT images, and electron-density map data.