Delineating vessels in an angiogram involves two methods: graph generation
and skeletonization. Generating a graph includes obtaining a digital image of an
angiogram, recognizing a first growth point within the image, and identifying region
boundary points around the growth point. The region boundary points are connected
to the first growth point, thereby creating edges of a graph. The boundary point
that has the greatest intensity is then selected as a second growth point, and
additional region boundary points around the second growth point are identified.
The additional region growth points are connected to the second growth point. The
region boundary point with the greatest intensity in the image is then selected
as a third growth point, and the method repeats until each point in the image is
connected to another point in the graph. The skeletonization of the graph begins
with recognizing a point in the graph as an endpoint of a vessel. This may be done
explicitly through manual or automatic selection of specific points. It may also
be done implicitly through a trimming process whereby graph branches of fewer than
a certain number of connected points are discarded. The endpoints in the remaining
branches are recognized as vessel endpoints. The skeletonization concludes with
display of the delineated vessels. This may be done by superimposing the vessels
in two or three dimensions over a conventional two-dimensional angiographic image
such as a maximum intensity projection (MIP).