A Novel Architecture and Analysis of Challenges for Combining Text and Image for Medical Image Retrieval

Abstract

The introduction of digital information storage and internet technologies has changed diagnostic radiology practices significantly. Radiological support systems including PACS (Picture Archiving and Communication System), SRS (Structured Reporting System) and image and text retrieval systems have been developed to assist radiologists. In this article we analyse the integration of image and text retrieval is promising for enhancing radiological teaching, research and diagnosis. Presentation of similar images as diagnostic aids has proven to be helpful in domains of radiology such as chest radiographs, lung CT and mammograms. We propose a diagnostic support system based on a combination of image and text retrieval. Our approach has the potential to increase the radiologist's confidence in his/her diagnosis which could translate to greater accuracy and reduced health care costs.