Three-dimensional Reconstruction Of The DSA Image Of The Neck Vein Based On The Combination Of Automatic Snake Model And Contour Rendering

In the field of medical imaging diagnosis,digital subtraction angiography(DSA)is the gold standard for clinical diagnosis of vascular diseases.At present,the widely used angiographic diagnosis and treatment is based on a large number of transcendental experience of clinicians to do perceptual assess of 2D vascular images.The appearance of 3D reconstruction of angiography provides a data based diagnosis and treatment model for clinical diagnosis and surgical guidance of angiographic images.However,there are still defects and problems in the clinical application of 3D reconstruction of DSA images at present,for example,the accuracy of feature matching is low,the fitting mode of central axis matching point is tedious,the reconstruction model lacks visual perception and so on.Based on the emphasis and difficulty of 3D reconstruction in the present stage,this paper proposes a method of combining automated Snake model and contour rendering for 3D reconstruction of cervical angiography vessels.This method solves the problem that the traditional Snake model must rely on prior knowledge to manually calibrate the control points and the lack of depth perception of the contrast image.The main methods are as follows:It has been put forward that extraction of cervical artery axis can be carried out by use of Gabor multi-scale filtering and Hessian matrix non suppression maximum value extraction algorithm.Also,it has been creatively pointed out that sift character matching for vascular axis of polar geometry relation is conducted based on double plane angrography Therefore,the automatic localization of contour control points is optimized for the Snake model during three dimensional reconstructions of the neck vessels.Research has been carried out for the contour rendering method of the reconstructed cervical vein.Specifically speaking,an Alpha contour rendering model is constructed based on depth buffer model.So,that has provided the visual perception model for clinicians’ and patients ’structure and deep perceptions of blood vessels.Hence,the conclusion can be drawn from the data analysis of the crucial results of the experiment.That is to say,in comparison with those of the traditional algorithm,the recall ratio of the now-adopted algorithm,which is based on the sift character matching of polar geometry suppression,is much closer to 1,and the robustness of it is higher,when matching is carried out for the vascular framework with rotation and angle transformation characteristics.Besides,the result can be gotten that the reconstruction error of the automated Snake model in this paper is 3 pixels less than that of the traditional Snake model,through the comparison of iterative errors of the Snake model.By evaluating the rendering results based on the perceptual evaluation of the Likert scale,the results of the rendering results in the depth,structure and color perception are higher than the traditional X-ray and reconstructed images.