Research On Establishment Of CTA Model Of Cardiac Coronary Based On Image Registration

With the rapid development of CAD(Computer Aided Diagnosis)technology,the importance of the system is higher and higher in clinical medicine.As an important part of CAD system,image registration plays an important role in the diagnosis of cardiovascular diseases.The location information of the coronary heart is an important basis for the diagnosis of cardiovascular diseases,therefore,how to make the combination of image registration algorithm and modeling method to obtain the direction information of the coronary region of interest and the coronary centerline,and making the automatic extraction and identification of the centerline of the three main branches of the coronary artery combined with the shortest path algorithm is a hot research topic currently.With the CTA(Computer Tomography Angiography)data taken as the research object,firstly,the 3D visualization technology is studied to make the 3D display of cardiac coronary CTA data with surface and volume rendering techniques;And then the image registration algorithm is analyzed in this thesis.Focusing on the problem of how to automatically segment the coronary line from the raw cardiac CTA data,the affine transformation,B-spline deformable transformation,image modeling and image fusion are analyzed and studied,and a method for establishing coronary CTA model based on image registration is proposed in this thesis to get the ROI(region of interest)and directional information of the coronary centerline;Finally,based on the prior model,a new method is proposed to full-automatically extract and recognize the centerlines of the major branches in the region of interest combined with minimal path algorithm.To sum up,the main contents are as follows:(1)In the process of 3D visualization,the surface rendering and volume rendering techniques were studied to realize the three-dimensional display of cardiac coronary CTA data.Based on this,a three-dimensional visualization software is developed using graphical user interface technology to implement some common functions in image processing: including rotation,zooming,setting the values of color and gradient,photo screenshots,and so on.De-noise processing of the experimental data is performed to prepare for subsequent registration experiments.(2)The three-dimensional image registration algorithm based on affine transformation and B-spline deformable transformation is studied,and the performance differences are studied and compared between them.The experimental data of the CTA model is performed on the affine transformation and B-spline deformable transformation separately,with mapping the experimental data to the same coordinate space,and the performance of the two algorithms is evaluated in this thesis.Based on this,an optimized registration algorithm is proposed to achieve the registration of the coronary heart.The Shape-based Averaging method is used to average the eight segmentation maps which mapped to the same coordinate space to build a CTA model with certain generalization ability.Based on this,the region of interest and directional information of the coronary centerline is obtained.(3)Based on the prior model,the position of the starting point is detected automatically by the learning-based algorithm in the region of interest of the CTA volume.And,the directional factor of the coronary artery is imported into the cost function of the minimal path,guiding the coronary centerline tracking process.Finally,the three major branches are automatically extracted and recognized in the region of interest of the coronary centerline.