Medical Image Retrieval: ISSR at CLEF 2009

This paper represents the first participation of the Institute of Statistical Studies and Research at Cairo University group in CLEF 2009-Medical image retrieval track. Our system uses Lemur toolkit for text retrieval. The main objective is to carry out retrieving medical image depending on associated image text. We experimented with different text features such as article title, image caption and the article paragraph(s) denoting to the image. We propose a simple and effective extraction method to find relevant paragraphs based on the structure of HTML files. Automatic translation of queries in different languages other than collection language is also experimented. In this paper the results of 9 runs are presented in order to compare retrieval based on different text features and the effect of stop word lists and the use of relevance feedback. Categories and subject descriptors H. Information Systems; H.3 Information Storage And Retrieval; H.3.1 Content Analysis and Indexing; H.3.4 Systems and Software;