Registration Method of Virtual Model and Real Bone Based on Feature Points for Surgical Navigation

摘要

The surgical navigation system needs to realize the registration between the patient's real bone and the virtual model of the patient's bone, and the accuracy of the registration directly affects the success or failure of the operation. This paper proposes a registration scheme based on feature points. Before the operation, the 3D-SIFT (Scale-invariant feature transform) algorithm was used to extract feature points on the virtual model. During the operation, the doctor used the NDI target pen to mark the corresponding feature points on the patient's bone. The registration of the feature points marked before and during the operation is realized by the ICP (Iterative Closest Point) algorithm, and then the registration between the patient's bone and the virtual model is realized. Experiments show that the average time for the registration scheme to achieve a registration is 8.64 seconds, and the average registration error is 2.9 voxels, which has high efficiency and accuracy.