Angiographic Image Segmentation And Reconstruction

Angiography is a means of imaging techniques that are widely used in the diagnosis of vascular disease. Magnetic resonance angiography (MRA) and CT angiography (CTA) is the primary selection for vascular imaging and vascular disease screening and diagnosis in clinical. In the post-processing of the angiography images, the vessel segmentation level of precision, is the key of vascular visualization, diagnosis of disease, treatment and evaluation, the virtual surgical and surgical guidance and the series of subsequent operation, but also the development bottleneck of medical image processing.Currently, medical image segmentation methods no less than thousands, but there is no one universal method can be applied to a wide variety of medical images. Choice of vessel segmentation methods is not only related to imaging modalities, but also links with different specific tissues and organs. In this paper, according to the differences of the anatomical structure and the angiography image characteristics of the blood vessel, we use different methods to deal with the blood vessels of different parts.For the image of the blood vessels of the abdomen, neck and other parts, the distribution of aorta is more concentrated, the segmentation is also relatively simple. So the segmentation method based on regional growth is selected and reconstructing the segmented results. To cerebral angiography image, the distribution of target is more discrete, and having so many branches, so the segmentation is more complex. In this paper, preprocessing the original image by many filtering methods, and comparative analysing the results of traditional segmentation methods, we propose a vessel segmentation method based on the local maximum between-class variance (Otsu). First, preprocessing the original image with anisotropic diffusion filtering. Then, determining a best size of the sub-block by calculating the rate of change of the image internal standard deviation between the different image sub-block size, the image is divided into a number of sub-image. And then, the sub-image is segmented using the Otsu algorithm, and merging divided sub-images to the final segmentation result. Finally, reconstructing segmentation results using the VTK (visualization Toolkit), and analyzing the results by comparison. Based on the medical image segmentation and registration tools package ITK (Insight Segmentation and Registration Toolkit) and VTK platform, we complete the angiographic image segmentation and3D reconstruction. and project reconstruction results into a two-dimensional image, which is compared with the manual segmentation template result, then evaluating them with simple two-dimensional image quality evaluation methods. The experimental results show that the segmentation method we selected can improve the recognition rate of vascular detail in cerebral angiography image, and obtain the better segmentation and reconstruction results.