Compression of Cardiac X-Ray Images Based on Acquisition Noise.pdf

Compression of Cardiac X-Ray Images Based on Acquisition Noise Frederik J. de Bruijn, Marc Schrijver, and Cornelis H. Slump University of Twente, Dept. of Electrical Engineering, Signals Systems – Network Theory P.O. Box 217, 7500 AE Enschede, The Netherlands Abstract The goal of this work is to illustrate the feasibility to use noise as a measure for data loss in compression of cardiac X-ray images. In lossy medical image compression, the requirements for the preservation of diagnostic integrity cannot be easily formulated in terms of a perceptual model. Especially since, in reality, human visual perception is dependent on numerous factors such as the viewing conditions and psycho-visual factors. In general, due to the low exposure, cardiac X-ray images tend to be relatively noisy. In the assumption that noise, in case it is not correlated with the signal, contains no diagnostic information, we propose a compression scheme, based on (overlapping) block transforms, exploiting the different spectral distributions of signal and noise. 1 INTRODUCTION 1.1 Perception- versus noise-based quantization An important issue in lossy compression of medical images is the risk of destroying diagnos- tically relevant information. Current lossy compression standards, such as JPEG and MPEG, are designed for conventional still-image and video display respectively. These algorithms are intended for viewing situations in which loss of image detail is not regarded as a loss of information to the observer, but is primarily assessed in an aesthetical sense. The quan- tization in conventional lossy image compression is mostly based on commonly accepted models of human visual perception. Some successful attempts have been reported of the application of perceptive criteria in lossy medical compression [1, 2, 3]. It is debatable, however, whether perceptive criteria are reliable for judging the preservation of diagnostic information. In this paper, we use a criterion for data loss which is n