The Research On Vascular Segmentation Algorithm Of Coronary CTA Images

Cardiovascular disease is very common and one of the diseases with the highest morbidity and mortality in the world today.With the development of medical standards and medical imaging technology,doctors can have a intuitive understanding of the cardiovascular disease through medical imaging.Nowadays,many imaging techniques for the diagnosis and treatment of coronary heart disease have been developed.Computed Tomography Angiography(CTA)has the advantages of low price,simple operation and non-invasive imaging.It is one of the most commonly used diagnostic methods for coronary heart disease.With the rapid development of computer-assisted diagnosis and treatment technology,clinical requirements for CTA image automation have been raised.In this paper,the coronary CTA volume data was used as the research object,and the research of coronary vessel segmentation was carried out from 2D and 3D,respectively,and the following research results were obtained:(1)From the research of 2D sequence image segmentation,this paper proposes an improved particle filter algorithm based on multi-features,which realizes the tracking and segmentation of vascular in coronary CTA sequence images,and render the 3D model of vascular by surface rendering algorithm.In the tracking algorithm,firstly,the bifurcated coronary vessels are tracked by the feature matching algorithm;secondly,the multi-features are combined and the particle resampling rules are improved to realize the tracking of small vascular with topological structure and positional changes;Finally,the center of the tracking target is used as a seed point to achieve sequence segmentation of vascular.Through experimental data analysis,the accuracy of coronary target tracking in this algorithm reaches97.84%,and compared with other tracking methods,the superiority of the algorithm in this paper is verified.(2)In order to realize automatic segmentation,from the research of 3D volume data segmentation,this paper proposes a 3D W-Net with adaptive weight loss function(3D W-Net With Adaptive Weighted Loss,AWL-W-Net).In this process,the W-Net network structuretraining segmentation model is established;Secondly,combined with the characteristics of coronary CTA images and W-Net network training results,the adaptive weight loss function layer is proposed,which enhances the network learning ability and achieves the goal of improving the segmentation accuracy of the vascular model and repairing the fracture of the3 D model.AWL-W-Net can complete the high-precision segmentation,and compare the results with W-Net and 3D U-Net network,which proves that AWL-W-Net network is not only improved the accuracy of segmentation,and the fracture of the model can be repaired,providing doctors with a high-precision 3D model of coronary vessels that is more in line with clinical needs.