Efficient geometric algorithms for determining motion and shape deformation of coronary vessels

Each year approximately 225,000 Americans die of vascular disease alone. For effective diagnosis and treatment of such diseases, accurate imaging of the complicated vascular structures is essential. Quantitative measures relating to motion and structural changes of vessels are extremely important for clinical diagnosis of such diseases. Current medical imaging techniques, however, cannot yet provide the necessary spatial/temporal resolution for such advanced analysis of the rapidly moving coronary vessels. Thus, a great deal of information between two consecutive image frames could be inherently lost. Furthermore, due to the low contrast in many images of coronary sequences, segmentation algorithms often fail to detect the exact vessel structures. In this paper, we propose geometric techniques to recover the motion and deformation of coronary vessels between two image frames. Our techniques can be used to increase the temporal resolution of a coronary sequence to a desired level by predicting the motion and shape-deformation of the vessel-structure. They can also be used to predict the structure of the vessel in a low contrast image to complement the results of a vessel-detection/segmentation algorithm; when used in conjunction with such a segmentation technique, it will help alleviate false positives and false negatives.