Contextual Region of Interest Based Medical Image Compression using Contextual Listless SPIHT Algorithm for Brain Images

摘要

Medical Imaging plays a major role in medical diagnosis. Storing these medical images and transmitting them is quite challenging. Due to the extensive use of medical images like CT and MR scan, the application of digital imaging technology in the medical domain has grown rapidly. These medical imagery are stored for a longer period for the continuous monitoring of the patients. So, the medical images need to be compressed to reduce the storage cost and for transmission without any loss. In this paper, a context based method called Contextual Listless Set Partitioning in Hierarchical Trees (CLSPIHT) algorithm for brain images is proposed to overcome this challenge. Here, the region containing the most inportant information for diagnosis purpose is referred as contextual region of interest. In this method, the Contextual Region of Interest(CROI) is encoded separately with a low compression rate ie, with high bpp and the Back Ground region(BG) is encoded with low bpp. Finally, the two regions are merged together to construct the output image. Our experimental results show that the proposed Contextual Listless SPIHT (CLSPIHT) yields very good image quality without any diagnostic loss. Compression performance parameters (Mean Square Error, Peak Signal to Noice Ratio, and Coefficient of Correlation) are improved by our method and it is compared with the other existing methods of JPEG2000,and the ROI based methods such as CSPIHT and CVQ on magnetic resonance images. As a result, it is found that our proposed algorithm gives better results and using this method, we can overcome the limitations in storage and transmission of medical images.