A Comparative Study on ROI-Based Lossy Compression Techniques for Compressing Medical Images

Medical image compression is an important area of research which aims at producing algorithms that reduce file size and at the same time maintain relevant diagnostic information. The main focus of this paper is to analyze techniques and find a compression scheme that can compress medical images quickly and reduce compression rate while maintaining a good level of visual quality. A Region Of Interset (ROI)-based approach is used to separate the important medical data and background data. The Block Truncation Coding (BTC) and Discrete Cosine Transformation (DCT) algorithms are used to code background region, while Embedded Zero-tree Wavelet (EZW) coding, Set Partition In Hierarchical Tree (SPIHT) Algorithm, Zero-Tree Entropy (ZTE) Coding Algorithm and Singular Value Decomposition (SVD) are used to compress the ROI region. Several experiments were conducted to analyze the algorithms based on compression ratio, decompressed image quality and speed.