A method for quantitatively estimating global displacement fields of coronary arterial vessel skeletons during cardiac cycles from X-ray coronary angiographic (CAG) image sequences is proposed. First, dynamic sequence of arterial lumen skeletons is semi-automatically reconstructed from a pair of angiographic image sequences acquired from two nearly orthogonal view angles covering one or several cardiac cycles. Then, displacement fields of 3D vessel skeletons at different cardiac phases are quantitatively estimated through searching optimal correspondences between skeletons of a same vessel branch at different time-points of image sequences with dynamic programming algorithm. The main advantage of this method is that possible errors introduced by calibration parameters of the imaging system are avoided and application of dynamic programming ensures low computation cost. Also, any a priori knowledge and model about cardiac and arterial dynamics is not needed and the same matching error function and similarity measurement can be used to estimate global displacement fields of vessel skeletons performing different kinds of motion. Validation experiments with computer-simulated data and clinically acquired image data are designed and results are given to demonstrate the accuracy and validity of the proposed method.
Read full abstract