A preliminary study is described relative to the possibility of using methods of automatic image interpretation in the field of biomedical images. These general methods are capable of recognizing characteristic shapes inside the image. In this paper, the structures are described and interpreted. These methods are capable of outlining shapes which are important from a clinical point of view--e.g., spine, vertebrae. They can also evaluate the shapes by plausible methods, to explain the meaning of the whole image. A feasibility study is described concerning the automatic recognition of vertebral contour in axial CT images. The results suggest that the employed techniques can be used in real diagnostic applications.