KIDS's evaluation in medical image retrieval task at ImageCLEF 2004

We describe our participation in the medical retrieval task of the ImageCLEF 2004. This task aims at finding images that are similar with respect to modality (CT, radiograph, MRI, and so on). We propose several image features, including color histogram, gray-spatial histogram, coherence moment, and gray correlogram, to facilitate the retrieval of similar images. The initial retrieval results are obtained via visual feature analysis. An automatic feedback mechanism clusters visually and textually similar images among these initial results to help refine the query. In this paper, we present the system used, focusing on novel and newly developed aspects. The evaluated result shows that the automatic feedback mechanism improves the precision by 15%.