| Abstract: | PURPOSETo validate the use of techniques of irreversible compression of images, which can be performed using a block-based discrete cosine transform technique as defined by the Joint Photographic Experts Group, before they can be used in clinical applications, by evaluating the effect of compression on the ability of observers to detect discrete white matter lesions on MR images of the brain.METHODSSixty T2- and intermediate-weighted spin-echo images were compressed with varying degrees of coefficient quantization with compression ratios from 1:1 to more than 40:1, randomized, and evaluated by three observers blinded to the degree of compression.RESULTSNo significant difference in the number of lesions detected was apparent until compression ratios reached 40:1, despite a significant subjective loss in perceived image quality at 20:1. Only small (< or = 5 mm) lesions were missed at the highest degree of compression. No significant differences were observed in the detection of confluent periventricular white matter disease at any degree of compression tested.CONCLUSIONSThe use of high degrees of irreversible compression of MR images may be acceptable for diagnostic tasks. |
