Articular Cartilage Segmentation Based on Radial Transformation

Osteoarthritis(OA) is one of the major causes of disability in the elderly population. The accurate segmentation of the articular knee cartilages from MR images is important to clinical diagnosis and treatment. In this paper, a semi-automatic segmentation method of knee cartilage based on radial transformation is proposed. A radial transformation is the process of generating a new image, called a radial image, by re-sampling the original image in a radial approach. The cartilage boundary is initially delineated on the radial image, and then transformed back to the original image. Due to the possible round-off errors and digitization effects, the cartilage boundary on the original image requires a slight deformation to become more accurate and complete. In general, the cartilage segmentation has been considered as a two- dimensional image processing challenge. However, the proposed method turns such a challenge into a one-dimensional job by incorporating the prior knowledge of cartilage anatomy into the radial transformation, and thus simplifies the segmentation problem.