Authors: Peter Hall, Richard Feltham, Trevor Fitzjohn
Source: GZipped PostScript (37kb); Adobe PDF (94kb)
An angiogram is an image of the blood vessels, and such images can be acquired via x-ray or magnetic resonance imaging. This paper discusses problems that arise when attempting quantitative x-ray angiography on a computer. Standard solutions are presented where they exist and we include proposals of our own. In particular, we focus upon the problem of reconstructing the three-dimensional network formed by blood vessels, from only a few angiograms. Our solution to this depends upon successful integration of methods drawn from computer graphics, computer vision, an anatomical model, and machine learning.