3-D image reconstruction of brain blood vessels from angiograms

This paper presents a reconstruction process to build the three-dimensional structure of cerebral vessels from a pair of DSA (digital subtraction angiogram) images. To segment the blood vessels from the background, not only the local thresholding but also global thresholding was used to bilevel the input image. Thinning process and the B-Spline curve are applied to extract and smooth the skeleton of the vessels. The skeleton is then tracked to find the branch points which will be used as the feature points for matching. For each feature point in the first image, we use the epipolar constraint and the property of feature points to find its correspondence on the other image. From the DSA machine structure and the correspondence information, the real 3-D position of the skeleton is calculated by least squares estimation and the reconstructed 3-D shading images for various viewing angles and light source directions are obtained using the Gouraud Shading method. Experimental results show that the proposed algorithm work well on our data and can be applied on similar data without intervention most of the time.