University of Pittsburgh Medical Center
Department of Pathology
Pittsburgh, Pennsylvania
Wen-Wei Chung, MD, PhD

Wen-Wei Chung, MD, PhD, Department of Pathology, University of Pittsburgh Medical Center
Charlie Hatton, Cambridge Technology Partner
James Woods, PhD, Cognitive Science Branch, National Library of Medicine
John Gilbertson, MD, Department of Pathology, University of Pittsburgh Medical Center
Michael Becich, MD, PhD, Department of Pathology, University of Pittsburgh Medical Center

ImageMed has evolved from a prototype developed in 1996 by Charlie Hatton and James Woods to a current working system. This project has been under a constant support from the US National Library of Medicine (NLM), the University of Pittsburgh Medical Center (UPMC), DFD Information Services, and the University of Illinois at Urbana/Champaign (UIUC). ImageMed allows clinicians and researchers to share content-based clinical images using output from the UMLS Metathesaurus at the National Library of Medicine.

ImageMed allows online image uploading to a centralized WWW server, indexing using a 1998 Metathesaurus keyword list, and enhanced search capabilities including search terms selected from a keyword list generated from Metathesaurus terms. The image database contains physical image files saved to the centralized WWW server, as well as links of image files to specific URLs from different providers. Currently, ImageMed contains a total of 766 images, including 11 prostate images from NLM, 132 images of general pathology and 47 images of cardiovascular pathology from UIUC, and 396 images of genitourinary pathology and 180 images of transplant pathology from the UPMC Department of Pathology online case database.

The development of a comprehensive distributed repository of pathology images will serve as a valuable resource for a wide range of health care and clinical research applications. The utility and success of such a repository is based on several factors. They include convenient user access to a facility for uploading and indexing images from local machines to a centralized image server, intuitive and powerful search capabilities, and the use of a uniform language set to allow for unambiguous indexing or identification of image contents. We are currently recruiting partner institutions and individuals to establish a comprehensive relational database of pathology images that may be searched using a variety of methods based on Metathesaurus output. The image upload, indexing, and search tools will eventually be made available to the health care community, and the image repository presented as a resource which can be shared and utilized via the WWW by all interested individuals.