Medical image retrieval based on low level feature and high level semantic feature

A medical image retrieval system combined of the low-level image feature and high-level semantic is used in the paper witch includes two main parts: image preprocessing and the machine learning. In the first part, feature tree structure is presented to reduce the semantic gap and in the latter part, a novel machine learning method based on SVM is presented to optimize the Network parameters by which improve the effect of semantic annotation and recognition rate. Preliminary test results form clinical images prove feasibility of the retrieval system and support the theory presented in the project.