Diagnostic Classification of Leg Radiographs

Medical imaging is a powerful and increasingly popular tool. This project addresses two major problems: data storage and transmission; and automated diagnosis and classification. We suggest that by addressing the latter issuue using advanced reasoning tools and clinical models (necessary in even the most deceptively obvious modailities), medical images (and subregions within them) can be categorised according to medical standards. Past experience has suggested that simple techniques fail to provide accurate results; and that to cope with the wide diversity of image information within any one medical modaility and subdomain (e.g. leg radiograophs) an adaptable system that combines evidence from multiple sources is required. The outcome from using such a system will provide a means for indexing large numbers of images. Furthermore, we suggest that the clinical classification of an image is of importance in categorisng image sub-regions and so providing a means by which differential information retaining compression techniques can be used, alleviating many problesm related to dtaa storage and image transmission. These problems need to be addressed as we move towards a technology dependent medical domain using Picture Archiving and Communication systems and Telereadiology techniques for diagnosis, conference and teaching purposes.