Efficient lossy to lossless medical image compression using integer wavelet transform and multiple subband decomposition

Abstract

Since medical imaging produce prohibitive amounts of data, efficient and low-complexity compression is necessary for storage and communication purposes. In this paper, a flexible coding algorithm called embedded multiple subband decomposition and set quadtree partitioning (EMSD-SQP) is presented based on integer wavelet transform (IWT). The presented method exploits three new coding strategies-multiple subbands decomposition (MSD), multiple subbands scaling (MSS) and fast quadtree partitioning (FQP). During transform, three high frequency subbands are secondly decomposed using IWT respectively for optimizing the transform coefficient distribution of high frequency subbands. Then, each subband is scaled according to its significance. The scaling factors are the integer powers of two. Finally, all image coefficients are encoded using fast quadtree partitioning scheme, which improves the lossy compression performance and reduces the memory demands. Simulation results for CT and MRI images show that the EMSD-SQP algorithm provides PSNR performance up to 4-6 dB better than SPIHT and SPECK using IWT. And the PSNR performance of EMSD-SQP has 0.4-0.8 dB better than SPIHT using Daubechies 9/7 discrete wavelet filters. Additionally, the lossless compression performance of the presented algorithm is quite competitive with other efficient compression method. © Springer-Verlag 2004.