Research Of Anterior Cruciate Ligament Rebuilding Based On Three‐dimensional Reconstruction Of Knee‐joint

Three dimension reconstruction is a classic research topic in the fields of Computer Vision and Medical Research. With the development of combining between medical imaging and computer technology in recent years, medical reconstruction has great demand and developmental potential in both theoretical research and applications. Anterior Cruciate Ligament(ACL) is one of the main structures which stabilize the knee, and is also the most vulnerable structure inside the knee. Currently, the main way to detect the damage of ACL relies on medical observation, in which external physical observation is easy to make large error, and interior observation under arthroscope can get precise result, but the minimally invasive operation cannot be accepted by all patients. Now, the most common way is based on three-dimensional reconstruction of CT/MRI scanning data, showing the interior structure. However, the developmental technologies in the application of biological tissue have not achieved the ideal results.Aiming at the questions mentioned above, this thesis further studies the current technologies of three-dimensional reconstruction based on images, and researches the hot spots of ACL reconstruction. A high-efficiency and low-cost binocular reconstruction algorithm is proposed in this thesis. This thesis breaks through the existing ACL reconstruction researching, and the proposed algorithms get better experimental results. The main constructions are as follow:Firstly, a new and high-efficiency binocular processing algorithm is proposed in this thesis, with adding the processing of epipolar rectification and changing the traditional feature-based matching model. The proposed algorithm can directly be applied to low-cost consumer cameras, and also can largely reduce the time consuming under the condition of having not negative effect on accuracy. The new proposed technology can provide doctors a better 3D model to observe external physical structure of patients’ knee.In addition, this thesis makes contribution to improvement on segmentation of soft tissue inside knee joint by reconstructing CT/MRI scanning data. Space model contrast clustering segmentation algorithm is proposed based on the basic processing of threshold segmentation and contour editor. The proposed algorithm can segment ACL from soft tissue better. Judgment of damage can be verified by measuring the physical parameters from the 3D model of ACL.Furthermore, a new algorithm to locate the anatomical position which is very important when making scheme of ACL rebuilding injure is proposed in this thesis. The new locating algorithm mainly depends on the maximum expected clustering, making the anatomical position be much closer to the reasonable position around the area of attachment point.Extensional experimental comparison and statistical evaluation demonstrate that the proposed binocular reconstruction algorithm achieves better success than the state-of-the-art binocular reconstruction technologies did, that the proposed space model contrast clustering segmentation algorithm makes the reconstructed ACL much clearer, which would be great help to disease diagnosis, that the new location algorithm can locate the reasonable anatomical position around the original attachment area.