Traditionally, Picture Archiving and Communication Systems (PACS) use textual-based retrieval, which have their limitations. General-purposed content-based image retrieval (CBIR) systems often do not perform well in medical images and are not integrated with PACS and Radiology Information System (RIS). In this presentation we design a CBIR system that is integrated with PACS and RIS, by using a user-supplied query image to retrieve similar images from PACS and get corresponding reports from RIS. We also employ ACR index for radiological diagnosis to reduce the search space and to provide meaningful results in our CBIR system. We use high resolution CT lung images as the test data. A key image is selected for each series, and after a radiologist delineates the pathology bearing region, local texture features as well as ACR indexes and series UID are stored in a CBIR server. Series UID can be used to retrieve images from PACS and to obtain corresponding reports from RIS. The system is a useful learning tool for radiology education and can provide valuable references for radiologists when a new case comes.

Date Published: 16 March 2006
PDF: 8 pages
Proc. SPIE 6145, Medical Imaging 2006: PACS and Imaging Informatics, 614515 (16 March 2006); doi: 10.1117/12.652671

Published in SPIE Proceedings Vol. 6145:
Medical Imaging 2006: PACS and Imaging Informatics
Steven C. Horii; Osman M. Ratib, Editor(s)