A Web Service for enabling Medical Image Retrieval Integrated into a Social Medical Image Sharing Platform

摘要

Content-based visual image access is in the process from a research domain towards real applications. So far, most image retrieval applications have been in one specialized domain such as lung CTs as diagnosis aid or for classification of general images based on anatomic region, modality, and view. This article describes the use of a content—based image retrieval system in connection with the medical image sharing platform MEDTING, so a data set with a veiy large variety. Similarity retrieval is possible for all cases of the social image sharing platform, so cases can be linked by either visual similarity or similarity in keywords. The visual retrieval search is based on the GIFT (GNU Image Finding Tool). The technology for updating the index with new images added by users employs RSS (Really Simple Syndication) feeds. The ARC (Advanced Resource Connector) middleware is used for the implementation of a web service for similarity retrieval, simplifying the integration of this sei-vice. Novelty of this article is the application/integration and image updating strategy. Retrieval methods themselves employ existing techniques that are all open source and can easily be reproduced.