Analysis And Research Of Geometrical 3-D Reconstruction Based On IVUS And CAG Images

Coronary heart disease(CHD)is a major disease that seriously jeopardizes the health of the world.If diagnosed as CHD,patients would be recommended by the most commonly measures of Coronary Angiography(CAG)and Intravascular Ultrasound(IVUS)for the diagnosis and treatment.CAG is regarded as the gold standard method for the current diagnosis of CHD.CAG images display the anatomical structure of coronary artery and the position and extent of the coronary artery stenosis clearly in the two-dimensional(2-D)projection image.IVUS is to introduce the ultrasound probe into the lumen of blood vessel to detect by a cardiac catheter.It could obtain the detailed anatomical information such as the luminal area,wall thickness,and plaque of the blood vessel.However,CAG is unable to provide structural information and extent of the vessel wall,ultrasound catheters cannot successfully pass through the severe lesion area due to the limitations of current catheter technology,and the spatial location of lesions in the coronary arteries cannot be obtained through IVUS.Based on the characteristics of CAG and IVUS,this paper combined IVUS images with CAG images to reconstruct three-dimensional blood vessel models for displaying coronary vessel details more accurately.The main research contents include that:1.Image preprocessing.For CAG images,the LoG operator with the artificial marker method were used to extract the edges of the blood vessels.Hessian matrixbased method was used to extract the blood vessel center line.For the IVUS images,the nonlinear diffusion method was used to perform the pre-processing.And then,the region growing-based optimization algorithm was proposed to extract the border of the Media-Adventitia based on the IVUS images grayscale and blood vessel boundary shape information.On the basis of the extraction of the Media-Adventitia boundary,the maximum class-variance threshold selection method was performed to extract the boundary of lumen.2.Reconstruction of three-dimensional centerline of blood vessels.It is proposed that the points coordinate on the three-dimensional(3-D)centerline of the blood vessel can be obtained by the method using the geometrical correspondence relationship of the blood vessels in the angiographic plane,and the smooth curve can be obtained by fitting the three-dimensional centerline of the blood vessel with the B-spline curve.Experimental results confirmed the effectiveness of the method.3.The localization and orientation of IVUS image sequence in the threedimensional centerline of blood vessel.The localization of IVUS image sequence was that the IVUS images were equidistantly and vertically placed along the 3-D centerline to fulfil the localization according to the best vertical plane method.The orientation method was proposed that the IVUS image was rotated at a certain angle and backprojected to the two angiographic planes to calculate the error.While the minimum error sum of the IVUS image back-projected to the two angiographic planes,the best orientation angle was the rotated angle,calculated by the minimum error sum.Experimental results confirmed the effectiveness of the method.This paper verified the accuracy of the 3-D model of vascular obtained by fusion of CAG images and IVUS images in the following aspects,the result on reconstruction of 3-D centerline,the result on fusion IVUS images with 3-D centerline.This study has great research value and practical significance for assisting clinical diagnosis and treating CHD.