Abstract
Tracking the motion of coronary arteries in cineangiograms is useful to describe the motion of heart wall and to examine the condition of coronary flow based on spatial and temporal distribution of the density of contrast material. Cineangiograms used for the motion analysis have poor contrast because the quantity of contrast material is forced to be limited in order to keep the heart beat as normal as possible. Moreover, examination of coronary flow necessitates analysis of the regression process of the contrast material.
This paper describes the digital processing methods for moving images to analyze the motion of arteries in such cineangiograms. Tracking methods are different for such distinct parts as branch points and curved corners and for nearly straight parts of arteries. This is because the former parts have low correlation with their neighbors, whereas the latter parts have high correlation.
The template matching method is used for the detection of corresponding distinct parts in adjacent frames. The prediction for the positions of corresponding points based on the results of tracking in previous frames is utilized for faster matching.
For tracking nearly straight parts, a circle is drawn with its center at a distinct point. Then a pair of peaks in differential distribution of the density along this arc are chosen as wall boundaries of an artery. Again, the prediction for the positions of wall boundaries based on the result of tracking in the previous frame is used for faster tracking.
The above methods are successful in tracking the motion of arteries with poor contrast cineangiograms and in drawing regression curves of contrast material at different parts on the arteries, from which the condition of blood flow can be examined.
This paper treats the motion of arteries in cineangiograms, that is, the motion in the two-dimensional projection plane. Three-dimensional motion can also be analyzed by applying above methods to biplane-type cineangiograms.
