Based the vascular - section of the two-plane orthogonal projection image reconstruction
|Keywords||Image reconstruction Cross section of blood vessel Image moment function Volume rendering Ray-casting algorithms Biplane orthogonal projections Projections of limited view angle Image reconstruction from projections|
Image reconstruction of the blood vessel cross section from biplane orthogonal projections is a representative image reconstruction problem with limit two orthogonal projection views and has important clinical application values in the diagnosis and therapy of all kinds of cardiovascular and cerebrovascular diseases. Biplane orthogonal angiography is a standard procedure for making the assessment of a patient’s actual vessel morphology as accurately as possible. After injection of a contrast dye into the ventricles, great arteries, or coronary arteries, the biplane angiography make time-equivalent projection images from two orthogonal directions. It allows morphological and functional imaging of selected anatomical structures with high temporal and spatial resolution. We mainly focus on the research on the approach of high precision blood vessel cross section reconstruction.The binary images of blood vessel cross section are developed, since the gray level of DSA (Digital Subtraction Angiography) image has no exact physical means in the image reconstruction. According the cross sections of healthy artery positions are elliptical and stenotic portions can have crescent, star or irregular shapes, we use elliptical, crescent, star and irregular shapes as our research model. Based on the binary images, the biplane orthogonal projections are precisely simulated using the ideal X-ray parallel-beam.Cross sectional reconstruction from two orthogonal projections is a typical ill-posed problem. According Uniqueness Theorem of image moment function, the binary matrix could be precisely reconstructed from its projections if all of its order moments are uniquely determined. In reconstructing this binary matrix from its horizontal and vertical projections, we calculate geometry moments of projections and invert them to the moment information of reconstructed blood vessel cross section. Because the projections is limited (only two views), in order to improve the reconstruction effectunder ill conditions such as sparse data, singular value decomposition is used to estimate the missing geometry moments, and then on this basis, calculate their Legendre moments which are orthogonal moments. In the meantime, we use intensity moment to convert the gray image to binary image during the iterative process. Once all missing moments are estimated; the image of cross section can be successfully reconstructed. In order to reduce the interfere of the noise, we also use high order and redundant intensity moment to reduce the noise and improve the precision of image reconstruction. Finally, we use the volume rendering method to reconstruct the 3D structure of blood vessel using 100 pieces of sliced sheet of blood vessel.