Abstract
Medical services today rely heavily on imaging technology, including X-ray computed tomography, magnetic resonance imaging, nuclear medicine examination, and etc. As these examinations become more and more common and more medical equipment, picture archiving and communication systems (PACS) have been proposed as tools for converting the image data in digital form and for handling the resulting amount of information. In order to make PACS work well, we need to compress the image data for efficient storage and transmission. The goal of image compression is to reduce the redundancy in the original image, since the adjacent pixel values are usually correlated. For medical image, lossless compression is preferred, because it does not degrade the image and can facilitate accurate diagnosis. So we use the integer wavelet for lossless compression. Compressing an image set with multi-dimension is very important in medical. By removing redundancy in the different dimension of images, we can acquire the smaller image data size than by conventional 2D method. Our study is to know the difference of redundancy between each dimension using correlative map. The larger redundancy is, the higher priority is. According to the priority, we can build a decomposition order from higher priority to lower priority. Conventional wavelet dyadic decomposition method is to remove the average redundancy existing in all dimensions of images and could not reveal the different redundancy in different dimension. So we have improved the wavelet decomposition method, which is called directional dyadic decomposition, to decrease more redundancy of images. And we use the 2D set partitioning in hierarchical trees (SPIHT) to code the pyramidal structure after directional dyadic decomposition. Compared with the dyadic decomposition, our directional dyadic decomposition with 2DSPIHT coding produced 10% increases in compression ratio. Besides, the compression ratio of our method is about two times higher than JPEG-LS.