AN ANALYTICAL LOOK AT THE EFFECTS OF COMPRESSION ON MEDICAL IMAGES

This article will take an analytical look at how lossy Joint Photograp hic Experts Group (JPEG) and wavelet image compression techniques affe ct medical image content. It begins with a brief explanation of how th e JPEG and wavelet algorithms work, and describes in general terms wha t effect they can have on image quality (removal of noise, blurring, a nd artifacts). It then focuses more specifically on medical image diag nostic content and explains why subtle pathologies, that may be diffic ult for the human eye to discern because of low contrast, are generall y very well preserved by these compression algorithms. By applying a w avelet decomposition to the whole image and to specific regions of int erest (ROI), and by understanding how the lossy quantization step atte nuates signals in those decomposition energy subbands, much can be lea rned about how tolerant various anatomical structures are to compressi on. High-frequency anatomical structures that have their energy repres ented by a few large coefficients (in the wavelet domain) will be well preserved, while, those structures with high frequency energy distrib uted over numerous smaller coefficients are the most vulnerable to com pression. Digitized films showing subtle chest nodules, a subtle stres s fracture, and CT and NIR images are used to show these results. Copy right (C) 1997 by W.B. Saunders Company.