We have shown that the 3-D binary representation of a tree-like object can be obtained from three mutually orthogonal projections. This is done by first backprojecting the binarized images from the three orthogonal directions and then iteratively removing artifacts in the backprojection. Three different algorithms have been developed and compared: the Lagrange multiplier algorithm (LMA), the conjugate gradient algorithm (CGA), and the minimum-voxel representation algorithm (MRA). We have found that MRA is superior than LMA and CGA because MRA exploits the continuity property of the tree structure and requires only binary instead of grayscale information from the input images. The above figure shows an example of the 3-view reconstruction of coronary arteries. The reconstruction based on MRA was accomplished in less than 20 seconds.
Relevant Journal Publications
Relevant Patents