A recurrent cooperative/competitive field for segmentation of magnetic resonance brain imagery

The gray-white decision network is introduced as an application of an on-center, off-surround recurrent cooperative/competitive network for segmentation of magnetic resonance imaging brain images. The three-layer dynamical system relaxes into a solution where each pixel is labeled as either gray matter, white matter, or 'other' matter by using raw input intensity, edge information, and neighbor interactions. This network is presented as an example of applying a recurrent cooperative/competitive field to a problem with multiple conflicting constraints. Simulations are presented and discussed.<
>