Abstract: It is a challenging task to build an efficient and effective retrieval system for medical images.A comprehensive survey of content-based medical image retrieval(CBMIR)techniques is presented in this paper.After a brief introduction of the CBMIR model,five related key techniques such as image feature extraction,feature selection and fusion,similarity matching,image prefiltering and relevance feedback,are deeply analyzed.Some latest applications are also discussed.Finally,some promising research directions about CBMIR are suggested.