Adaptive simultaneous coronary border detection: a method for accurate analysis of small diameter vessels

Accurate automated identification of coronary borders is difficult to achieve in vessels smaller than 1 mm. In diameter using existing techniques. The authors new approach uses a two-stage automated adaption of edge operator parameters to match the local lumen diameter. In the first stage, approximate lumen diameters are detected using a half-resolution simultaneous coronary border detection algorithm. At each point along the vessel, the approximate diameter is used to select an optimal edge operator for the second full-resolution border detection stage. The method was tested in a set of 72 segments of 9 angiographic phantom vessels with diameters ranging from 0.46 mm to 4.14 mm. Performance of the adaptive method was compared to performance of a conventional coronary border detection, method. Adaptive simultaneous coronary border detection significantly reduced diameter estimation errors in comparison to conventional border detection (p<0.001). Adaptive coronary border detection substantially improves border detection accuracy in small diameter phantom vessels