Report on the CLEF Experiment: Combining Image and Multilingual Search for Medical Image Retrieval.
Henning Müller, Antoine Geissbühler, Patrick Ruch · CLEF (Working Notes) · 2004
This article describes the technologies used for the various runs submitted by the University of Geneva in the context of the 2004 imageCLEF competition. As our expertise is mainly in the field of medical image retrieval, we will concentrate most of our effort on the medical image retrieval task. Described are the runs that were submitted by our group including technical details for each of the single runs and a short explication of the obtained results, also compared with the results of submissions from other research groups. We will also describe the problems encountered with respect to optimising the system and especially with respect to finding a balance between weighting the textual and visual features for retrieval. A much better balance seems possible when using some training data for optimisation and with the relevance judgements being available for a control of the respective retrieval quality. The results show that relevance feedback is extremely important for optimal results. Query expansion with visual features only gives minimal changes in result quality. If textual features are added in the automatic query expansion, then the results improve significantly. Visual and textual results combined deliver the best results.