| Region-based coding schemes, employed in JPEG-2000 and MPEG-4 standards, are promising for medical image compression. In most medical imaging applications, the diagnostically significant information is localized over multiple, arbitrary shape regions which require high quality coding and the rest of the image might be encoded with lower fidelity. Motivated by this fact, in this thesis, we investigate multi-region multi-quality (MRMQ) coding of medical images. Proposed methods differ from existing region-of-interest (ROI) coding schemes in such a way that maximally decimated wavelet transforms are used for representation of regions rather than the identification of the region information in the transform domain. Two new MRMQ coders based on SPIHT, JPEG-2000 and EZBC algorithms are presented. These coders offer superior rate-distortion performance with additional features such as distortion scalability, spatial scalability and flexibility in precise rate control. Furthermore, a novel scheme which addresses the region size sensitivity problem of tree-based wavelet coders is proposed. Experimental results verify the strength of the proposed MRMQ coders in the utilization of the available bit rate. |
