The brain CT image database is the result of the efforts made at the N:.161 Center Hospital for accumiely diagnosing encephalic illness. Hon. to prevent the man-made error and iniproye tlie interpret level for brain CT images. is the purpose of this database. so the content-based seinantic brain CT image retrieval process become important. We adopt some methods in order to ovcrcomne the semantic gap between low-Ievel visual features with high-level semantic features. Finally. Tve apply a new complex similarity query algorithm to fulfil1 the iinage tetrieval.