The Study Of Compression Method For Medical Image In PACS

In recent years, the demand for PACS(Picture Archiving and Communication System) is increasing with the establishment of digital hospitals. At the same time, the widespread application of medical devices produces the enormous amount of medical image data, which bring a serious challenge to the limited storage space and transfer bandwidth, Therefore, the study of medical image compression technology is of great importance.The survey shows that most of the traditional medical image compression methods are based on wavelet transform. Theoretically, wavelet has limited directions, and only can represents singular points of images accurately compared with the linear or planar singular, while most of the lesions in the medical image are the linear or planar singular. Because of the wavelet’s limitations in high-dimensional data representation, the MSSIM(Mean Structural SIMilarity) between the original image and the reconstructed image obtained by the medical image compression methods which are based on wavelet is lower, a new compression algorithm for medical images is proposed based on the shearlet transform to solve this problem. Moreover, on the basis of medical images’characteristics and the need of clinical application,a mixed medical image compression method, which makes lossless compression and lossy compression to the lesion region and background respectively, is proposed to apply to PACS. the following is some key research points of the paper:1. Shearlet has a lot of better performance than wavelet, which makes the shearlet transform is more suitable for image compression. This paper has a comprehensive study on the structure of the shearlet and the discretization and its’ related properties through theoretical analysis and experimental simulation. Experimental results show that shearlet transform has good sparseness, the localization of direction and low time complexity. All of these laid the foundation for the use of shearlet in medical image compression.2. The compression scheme proposed in this paper involves several key questions:the choice of basis function, decomposition level and coding method. It is Obviously that different basis function, decomposition leve and coding method will affect the compression effect directly, experimental results show that the proposed compression scheme has a better compression effect with the db4wavelet base,3decomposition level, and Huffman coding.3. The conflict management of the high compression ratio and the high quality of reconstructed image in PACS. Because of the significance of the lesion area in medical image diagnosis, its’reconstructed image must has good quality. In this paper, we proposed a medical image compression method based on ROI(ROI, Region of interest). The experimental results show that the proposed method, with configurable reconstructed images’and compression ratio, adapts to medical image compression in PACS.There are a lot of problems to be improved in the medical image compression scheme proposed in this paper, such as the optimization of the algorithm, the automatical selection of ROI,etl, These will be discussed in our future research.