KIDS (knowledge-based diagnosis system)-a specialized architecture

The design of architectures is presented for knowledge-based image diagnosis systems (KIDS). Two basic constructs (knowledge sources and knowledge processors) are first presented as modeling the capabilities involved in the tasks of problem modeling and solving. These constructs are then specialized to fit the image-interpretation problem. The architecture is conceived as a network of agents, each responsible for a well-delimited task. The agents cooperate in solving the problem at hand. It is obtained by distributing the knowledge, expertise, and control abilities. Examples are provided in the field of biomedicine (cyto- and histopathology)