Progressive medical image coding using set of hierarchical trees

A new coding method to exploit intra-subband and inter-subband correlation of the medical images in the transform domain is presented. Subbands corresponding to each direction are combined together and treated as a set of hierarchical trees. If a set of hierarchical trees becomes significant, it is divided into four sets and each set is processed separately. The image is encoded progressively and the user can stop decoding at any instant, depending upon the extent of clarity required. Experiments are conducted on CT images. The results show a significant improvement in terms of bit-rate for the required peak signal to noise ratio (PSNR) as compared to both progressive and region based image coding methods.