Automatic edge detection and diameter calculation of coronary arteries using DSA images

An automatic edge detection method based on a one-dimensional derivative and a diameter-calculation algorithm for stenosis quantification in DSA (digital subtraction angiography) images are presented. The algorithm is tested using both phantom and clinical data. The percent stenosis obtained from the algorithm are compared to the results obtained from caliper measurements on the projected cine images, agreeing within 10%. The algorithm worked well up to 75% stenosis. Above that level the algorithm either gave smaller results or failed completely by locking into a spurious channel next to the stenosed section of the vessel