Tracking vascular motion in X-ray image sequences with Kalman snakes

The quantification of blood flow has received much attention from researchers investigating atherosclerotic disease, a major source of heart attacks and strokes in the US. The authors are interested in tracking the coronary vessel motion in order to allow for successive angiographic image pairs to be brought into registration. Once this motion is measured, it can be subtracted out, allowing for more accurate measurement of flow in the coronary arteries directly from angiographic images. Here, the authors propose a vessel tracking scheme based on the Kalman-filter which uses B-spline models of the coronary vasculature in estimating their position and velocity over time. Measurements of vessel positions are made using both sum-of-squared-differences, and minimum intensity of image masks, and the measurements are combined to give robust estimates of the position and velocity of B-spline control points over time. Results are shown for two sequences of the motion of the right coronary artery