DIFFUSIVE EDGE ENHANCEMENT IN IMAGING

A quantitative theory of the effect of diffusion on images of a bounded region in the fast- and slow-exchange regimes is presented. In the fast-exchange regime, the image is motionally narrowed to a simple peaked shape, independent of the shape of the object. The peak is Gaussian for phase encoding and Lorentzian for frequency encoding. In the slow-exchange regime, it is shown that edge enhancement of images is expected, for both phase and frequency encoding. For frequency encoding, the shape of the edge peaks is Lorentzian at long echo times and an ''asymmetric Lorentzian'' for short echo times. Some features of the edge peaks for phase encoding can be understood in terms of simple properties of random walks. (C) 1995 Academic Press, Inc.