The use of digital imaging has been applied throughout various radiology services since CT was developed. Both economics and efficiency of data storage and transmission have been demanded by the medical imaging community. A new decomposition method using an image splitting and gray value remapping method, followed by alternate value contour coding for both highly correlated portion and edges is proposed for radiological image compression in this application. Depending upon the image quality (noise level of the acquisition device), this error-free compression technique can improve the compression efficiency by approximate 30-50% with respect to the conventional DPCM/Huffman coding. The practical compression ratios are varied from 3.2:1 to 4:1 and the true storage saving ratio is 5:1. In this SBIR project, we would like to (1) explore several artificial intelligence techniques for error-free radiological image compression, (2) evaluate the proposed compression method to various type of medical images, and (3) construct parameters and structure the design of compression and decompression modules. This project would eventually lead to an efficient data storage and a rapid image communication in medical information applications.
