Computer aids for decision-making in radiology

A key aim of work in medical informatics is to help clinicians make use of a super abundance of information. Decision support systems assist by organising the selection and presentation of information. We argue that, since a variety of computerised information sources are now conceivable, tools combining different kinds of information source are required. We have developed a generic architecture in which a model of decision making and of image interpretation provides the framework for combining symbolic knowledge and information derived from digital image processing. We have identified four kinds of knowledge source used in decision aids for radiological image interpretation: image databases; numerical decision aids; expert systems; and image processing.