Content-based medical image retrieval system using feature decomposition of medical images

Abstract

The use of accumulated medical images has been limited, even though the number of medical images stored in hospitals is growing faster than ever. This is because existing content-based medical image retrieval systems typically require example images to extract the query vectors. However, it is not always possible to prepare example images on demand. Here, we present a novel medical image retrieval system based on the feature decomposition of medical images. By decomposing the features of medical images into normal and abnormal features, our medical image retrieval system enables users to overcome previous challenges, including image retrieval without using example images. Our image retrieval system can expand the utility of medical image databases.