This work starts from the premise that, given the current considerable growth in medical imaging, there is a need to develop a method that allows the information thus gathered to be used to its optimum - not only as a separate data set but also within the context of other related data. From this has grown the concept of a visualization tool which aids the visual comprehension of data present in an image by using information both internal and external to it. As a result, key medical features should be identified, labelled and presented in a clear and meaningful manner. The development of the visualization tool has been achieved through the use of blackboard-based expert system. As well as providing a method for integrating the different models used, the blackboard shell has enabled all aspects of the visualization process to be centrally supervised using a powerful and flexible control mechanism that permits both goal directed and data driven behaviour within the system. The modular approach that has been applied permits the model-based processes of feature recognition to be developed as quasi-independent systems. Two feature recognition models have been developed and these are interfaced to the rest of the tool through a set of feature dependent experts that contain knowledge about how and when to use these models to their optimum. In addition, further modification to the prototype shell used has permitted the development and application of a feature sensitive search strategy. All components of the visualization tool have been tested separately and as a whole using real medical image data from a relatively low resolution source and have been proved to work. The regions and features information applied proved the viability of the overall-performance of the knowledge based feature models and allowed the results to be visually presented in a concise and original manner that provided additional information to an image without loss of the original information.