Research Of DICOM-based Medical Image Compression

With the development of digital medical devices and Internet, people become to concern PACS which mainly involve medical image format and compression. On the standard of medical image format, DICOM 3.0 was built by NEMA; on the compression of medical image, lossless compression and lossy compression are introduced to PACS. By lossless compression, the image has high quality but lower compression ratio which is difficult to achieve the requirement of storage and transmission. By lossless methods, it has high compression ratio but low quality of image, especially the regions of pathological changes, which make doctors diagnose imprecisely. People begin the research of ROI-based (region of interest) image compression from 1990s, but it is still on the first stage. The paper granted by Beijing Hospital discusses a new compression idea that has different quality in different regions based on ROI which solves the problem between storage and quality. The main contributions can be summarized as follows:On the standard of DICOM, this paper analyzes the storage service of DICOM3.0, and achieves the read, show and compression of medical image.On image compression, this paper discusses the basic theory of wavelet transform and constructs IWT (integer wavelet transform) using a new generation wavelet technology—lifting scheme. In addition, the paper analyzes deeply EZW (Embedded Zero Tree Wavelet) coding and SPIHT (Set Partitioning in Hierarchical Trees) on which we construct the algorithms of coding and decoding on medical image with ROI that is the key technology of the whole system, which solves the problem between compression ratio and image quality, that is, the algorithm improve the compression ratio and image quality in ROI. Hence this algorithm achieves the dual requirements of medical image storage and doctor's diagnosis. The experiments support the validity of our algorithm.