Research On Data Fusion Based On Image Segmentation Of IVUS And 3D Reconstruction Of CAG

Atherosclerosis which is a common disease in coronary artery disease is one of the primary human health hazards. It has two major clinical diagnostic methods: X-ray coronary angiography and intravascular ultrasound images. X-ray coronary angiography images can make an objective diagnosis for the structure of coronary artery and vascular stenosis. So we can obtain the three-dimensional spatial structure skeleton for direct observation about the coronary morphology through 3D reconstruction. Intravascular ultrasound can not only accurately measure the parameters of the lumen diameter and area, but also find the early coronary atherosclerotic plaque. So we can detect the eccentricity, wall thickness and the composition about plaque. We can make the advantages of both methods complement each other, through the data fusion between CAG and IVUS. We can obtain the solid model of coronary artery. At the same time, we can not only observe the spatial position of vessels, but also get the quantitative data of plaque for interested segment to help clinical diagnosis and treatment.The main contents of this dissertation include five parts as follows:1. Segmentation of lumen and media-adventitia border in intravascular ultrasound images based on improved Snake modelWe firstly use a spatio-temporal filter to reduce the effect of speckles and enhance the images. Then, we ameliorate the internal energy and design the image energy for reducing the sensitivity. Finally, through experiment we improve the local segmentation. At the same time, we give a optimization calculation for the energy function of Snake by greedy method.2. 3D reconstruction and optimization for catheter and vessel centerline in interested coronary arteryThis paper proposes a method to optimize the 3D reconstruction of interested vessels: optimization of matched points and GT. Through setting six pairs of initial matched points by chord formula, designing the energy function for matched points optimization, we can solve it using dynamic programming algorithm.3. Position calculation for IVUS image sequences located in 3D catheter The methods are presented for IVUS image sequences to locate in 3D catheter. The point sequences of catheter are smoothed by B-spline curve fitting. We figure out the cut vector by the method of optimal vertical plane. Then, we can realize the location of IVUS image sequences in 3D catheter through the coordinate transformation.4. Oriention calculation for IVUS image sequences located in 3D catheterThe methods are presented for IVUS image sequences to determine the orientation in 3D catheter. Relative twists are figured out by geometrical model of spatial curve between the adjacent frames. The eccentric angles can be figured out by the spatial relationship between matched points. Finally, we estimate the absolute orientation of IVUS image sequences in 3D catheter through quaternion and complex matrix translation respectively.5. Experiment of data fusion between the IVUS image sequences and the coronary angiogramThe data fusion between the simulated, clinical IVUS images and coronary angiogram are experimented respectively, and the accuracy of the method is validated using error analysis.