The diagnostic value of CT (Computed Tomography) checking for encephalic illness is affirmative. For clinical doctors, they are in urgent need of a good approach for this monomodality medical image fusion at an acceptable accuracy, in order to obtain some visual comparison about a patient in normal and pathologic conditions, tracing the development of focus, determining the regimen and so on. Thus is also the purpose of this paper. The usual method is merging images at pixel-level or feature-level. In this paper, we develop a semantic-level fusion technique that is matched with semantic descriptions associated to images. Content-based semantic information can be used on image segmentation and similarity matching image retrieval through prior-knowledge support. Then we apply a weighted complex similarity retrieval algorithm (WK-NN) to implement. Finally, the integrated images with semantic information are presented.