JISC

This start-up repository project will develop a warehouse of medical images and facilitate effective online retrieval tools in the institutional web site to complement the existing online e-leaning and teaching system OASISplus, also known as Blackboard Vista. Although OASIS+, can upload images, it is the way in which images are indexed and thereby retrieved effectively that poses a challenge.

Traditionally, indexing and retrieval of images is based upon textual annotations manually extracted from an image. To compensate the short comes of non-scalability and subjectivity of a text-based system, content-based image retrieval systems (CBIR) came along by querying the collections using a sample image. These text-based and CBIR systems are usually designed to work independently with few existing systems providing a retrieval facility accommodating metadata to allow retrieval to be performed via either text or query sample, or even more challenging both, especially in the medical domain where several features may be irrelevant, such as colour .

The repository, MIRAGE, is a subject-based (for medical research) and will cater for both text-based and content-based retrieval with a potential of semantic retrieval. The chosen platform is GNU GIFT, an open source that has been adapted in the community of archiving medical images. This project will develop an infrastructure interfacing with OASIS+ and services of the repository to encourage greater use of the repository, to the benefit of the University and community.

MIRAGE is structured into 2 stages. Stage 1 will deposit images into the repository with meaningful (semantic) tags, which allows multi-modal retrieval to take place in Stage 2. The main focus of the development will be the ingestion of the repository and the design of the infrastructure of an interface between MIRAGE and OASIS. The developed tools will be designed to be compatible with and to contribute to the JISC e-Framework. 

Lead Institution

• Middlesex University