Computer Aided Diagnosis (CAD) systems, as secondopinion assistance to radiologists, are being developed. However, the lack of semantic information at the image processing stage is expected to limit the ultimate performance of such CAD systems. A solution can be provided by a specific medical domain terminology, which contains several concepts that describe features of abnormalities and the relationships (properties) between these concepts; as such is well readable by both human beings (radiologists) and machine agents (CAD systems). Here we investigate a specific ontology and provide a clear evaluation based on string and Description Logic (DL). The results provide a clear understanding of semantic medical image analysis with which CAD systems are expected to improve their performance.