Vessel Segmentation and Stenosis Quantification from Coronary X-Ray Angiograms

Angiography is the gold standard for diagnosis and interventional treatment of vascular pathologies, especially for stenosis, in many hospitals around the world. Still, the physicians complain of visual burdens due to its rather low spatial resolution and artefacts. Fluoroscopic angiography series from the dataset are obtained with standard clinical protocol. In the following, there is proposed an algorithm for vessel segmentation and edge detection. It is related to gradient operator applied on a pre-processed image with the Frangi’s vesselness filtering for removing the equipment acquisition noises, followed by morphological operations for removing the spurs and adaptive threholding. The contour tracking along the vessel is done using Dijkstra’s smallest path algorithm. The severity of the stenosis for a vessel segment can be assessed visually by a medical imagistic expert. The angiograph software can provide a graphic containing on the X axis the vessel segment length and on Y axis its corresponding cross-sectional areas. More objectively, the percentage of area stenosis can be computed.