Research And Implementation On An Auto 3D Articular Cartilage Segmentation

With the increasing deterioration of aging population all over the world, the threat of osteoarthrosis(OA) to human health is becoming increasingly prominent, too. To diagnosis the level of osteoarthrosis, clinically, doctors observe the medical image sequence of disease region and estimate the shape, thickness and volume of articular cartilage by naked eye. The development of modern science and imaging technology has brought great potential for medical image processing, and makes it possible for us to use computer as auxiliary in the diagnosis of osteoarthrosis. In this paper, we propose a compound method of 3D cartilage segmentation based on the mature algorithms of former researchers, using the Hessian matrix and 3D level-set algorithms to extract cartilage from 3D images. Reconstruct the extracted cartilage in 3D, in order to show the shape and structure information of cartilage.The main works of this paper are:1) Using the eigenvalue of Hessian matrix to enhance sheet structure in 3D image, thus extract sheet structure of certain sizes. Then, we combine the intensity and connectedness information of image, to remove the none-cartilage data from the extracted sheet structure and get the rough cartilage contour.2) The contour of cartilage extracted in the earlier chapter is not accurate. Level set algorithm has an excellent effect on the contour evolution of 3D space, so, after considering several different level set algorithms, we introduce a 3D level-set algorithm based on the minimization of region-scalable fitting energy. Using this algorithm, we refine the former result and get the precise contours of cartilage tissue. To evaluate the result of our method, we use the MR image set of articulation as experience data, Compared with the handmade segmentation of experienced doctors, which is considered as golden standard.3) We design a 3D cartilage segmentation system with visual interactive interface using VC++ and QT. Using the algorithm described in the previous section to segment cartilage. Reconstruct the extracted cartilage in 3D using VTK. Using the volume rendering algorithm in reconstructing the initial 3D medical image data, and surface rendering algorithm in reconstructing extracted cartilage. Our system show the arthrosis and extracted cartilage in 3D space visually and clearly, with the main purpose of helping the diagnosis of osteoarthrosis.