The Digital Imaging and Communications in Medicine (DICOM) Standard was developed by National Electrical Manufacturers Association (NEMA) and the American College of Radiology (ACR) for medical image archiving and communication. The Standard is widely used in medicine now. But it does not define image semantic content sufficiently, so there is a difficulty in the implementation based on semantic content. In this article are discussed the semantic hierarchy structure in medical image and the method to describe the image content used in the DICOM Standard. We point out the limitations of DICOM in image semantic content description, and then set forth a way to extend the semantic content description by using DICOM private attributes.