An Intuitive Annotation Toolbox for Open-source Imaging Diagnosis

Gemo, Monica [UCL] Kieffer, Suzanne [UCL] Gouze, Annabelle [UCL] Lawson, Jean-Yves Lionel [UCL] Macq, Benoît [UCL] Niang, Mamadou [] SCHOBBENS, PIERRE-YVES [UCL] Mazy, Gilbert [UCL] Debande, Benoît [UCL] Farsi, Sidi Mohamed [] et al.[show all]

Multimedia semantic processing such as semantics extraction from medical images is considered a key technology to develop next generation computer aided diagnostic systems. However, the introduction of content technologies in clinical settings requires effective reliable tools for image interpretation. This paper presents a clinical imaging annotation system offering a composite open-source toolbox for image analysis, intuitive multimodal interaction and data storage. The proposed system aims at more consistent and efficient mark-up of digital images as an aid in diagnostic tasks and as a complementary tool to telemedicine. The annotations include image extracted features and user-defined subjective characterisation. All resulting annotated datasets are stored in a knowledge-based repository that facilitates multimedia artefacts retrieval (e.g. intelligent queries of similar cases providing complementary hints to the remote expert advice! in telemedicine). Annotation functionalities illustrate the interpretation of breast cancer screening mammograms. However, the solution outlined can be easily applied to other clinical domains.