A new method for structure recognition in unsubtracted digital angiograms

Existing methods for automatically finding arteries in coronary angiograms rely on preprocessing. The authors develop a structure recognition system which is not sensitive to variations in image quality. This system utilizes a probabilistic contextual segmentation technique which performs image segmentation without preprocessing. This approach, the deformable template matcher, combines prior knowledge of the arterial tree, encoded as mathematical templates, with a stochastic deformation process described by a hidden Markov model. An introduction to the technique is presented along with recent enhancements of the structure recognition system