Lossless Medical Image Compression Using Motion Estimation and JPEG-LS Coding

• Main Authors: Fu-Sheng Ke, 柯福生
• Other Authors: Shaou-Gang Miaou
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/84681554234695492823

碩士 === 中原大學 === 電子工程研究所 === 94 === As enormous amount of medical images is under digitization recently, medical image compression has become a hot topic for study. One of the digitized sources is from the commercially available capsule endoscopy, which takes two pictures per second. These images occupy 10.5 GB of data space in eight hours. If the latest capsule endoscopy, which takes four pictures per second, is used the required storage space is even larger. With the advancement of technology development, the charge for the capsule endoscopy treatment is expected to be less expensive, and more and more patients can afford it. Consequently a good resolution to the issue of storing such images is more and more important. To avoid the lawsuit in medical disputes, compression of medical images is better performed losslessly. Although lossless JPEG-LS has good performance on compressing capsule endoscope images, it only compresses a single picture with intra coding and does not utilize the temporal information in adjacent pictures. Therefore, this thesis proposes a method that combines motion vectors and JPEG-LS to enhance the compression performance of basic JPEG-LS. In the experiment, we have six capsule endoscope image sequences for testing, and each sequence consists of 100 images. Experiment results show that the proposed scheme achieves a compression gain from 4% to 31.6% than the method of using JPEG-LS directly. A further experiment attempts to compress the MRI and natural images. The compression results also show very good performance for MRI images, but no obvious improvement for natural images.