Limits of lossless compression of digital cine angiograms: Markov models

Abstract

Presents a novel method to estimate the limit of lossless compression of digital cine angiograms. Three Markov models are used, one explores temporal dependency, one explores spatial dependency, and one explores both temporal and spatial dependencies. The models are from simple progressively towards complex. They have used 256, 256/spl times/256, and 256/spl times/256/spl times/256 contexts respectively. Five digital cine angiograms collected from both Siemens Hicore and Digitron digital cine angiographic imaging devices have been used for the test. The results have shown that the highest lossless compression ratio could approach to 6:1, which is slightly less than the limit of visually lossless JPEG compression ratio reported.