Study On Image Correction & Location Technique Of C-Arm

The surgical operations navigation which is based on C- arm x Ray (referred as C-arm) images is one of the hot topics within area of utilizing computers to assist in surgical operations. Though Image Intensifier c-arm is well known for its low cost and widely usage, C-arm image cannot orientate and navigate location correctly, due the unavoidable image distortion. In this paper it did researches on correcting the distortion of C-Arm X-Ray projection images, calibrating C-arm image models and how to utilize C-arm to locate a target in space etc., which are important for civilizing C-arm surgery navigation in practice from both theory and practical value's perspective.The main tasks in this paper are as follows:Firstly, it analyzes the type of C-arm image distortion and the causes. In order to correct the image distortion, in this paper it suggests applying Moving Least-Squares (MLS) to increase virtual makers which could eventually improve the correcting accuracy. The MLS is feasible based on a feasibility analysis that has been conducted in this research. By comparing the correcting results of ideal models created by pinhole image projection method, isometric method, similarity method etc, along with model theory and experimental effect, pinhole image projection method is finally standing out as the method, which will be employed in this research, to generate the idealization of an ideal model. The pinhole image projection method improves the effect of correcting in whole.Then, in order to building an ideal model, in this paper it designs a new double layered correcting target based on pinhole image projection model. Additionally, complying with the features of projection images, this paper still suggests using morphological method to process images. Further, by analyzing the markers'regular pattern of C-arm projection images, auto-identification and extraction of coordinate information of different flat markers on the projection images are achieved. By sorting markers, image flat points are able to match with spatial point. Then according to coordinate extraction experiments on C-arm projection image from different angles, and the comparison of differences with the manual extraction result, this paper proofs the reliability, accuracy, and timeliness of the extraction techniques which applies in the experiments could be achieved.Thirdly, in order to achieve the space positioning, and build foundation for the following 3D -rebuilding, this paper recommends using Tsai camera mathematical model to process C-arm image system marking, which could reduce marking errors. To correct image accurately, this paper suggests adopting 8 points normalization method, which would achieve the correspondence of spatial points of projection image from different angle of C-arm. Then, by utilizing the epipolar Constraint panel which is specially designed for this research to verify the correspondence level of the spatial points, and along with the distance measurement, this paper testifies the accuracy.Lastly, all the techniques that have been researched in this paper are testified by clinic solid tests, which proof a good systematic achievement. It testifies the correcting technique's feasibility on animal speciums of dogs, and the reliability of the space positioning technique on human brain image experiments, which basically all reach the system's requirements and provide the foundation for further performance optimization.