Using an approach inspired by biological neural networks underlying the pre-processing stages of the mammalian visual system, we are developing computer-processing paradigms for edge detection and image segmentation. Simulations are carried out on a network of interconnected """"""""neurons,"""""""" to which pixel intensities associated with a 2-D image serve as inputs. The final activation state of the network represents the outlines of contrast features within the image. We are applying the imageprocessing techniques developed in this work to the analysis of nuclear medical images.