Shape-based approach for pointwise tracking of contours

For several applications in medicine, it is fundamental to determine the temporal tracking of structure points, such as the apex of the heart. This tracking can provide important information about the effective movement and contraction of muscles. This paper presents an alternative methodology for the point-wise tracking of contours based on intermediary contours. Given two closed contours [c(0) and c(t)] derived from two consecutive image frames, the basic idea is to estimate intermediary contours based on shape and then track prominent points of c(0) up to c(t), considering all contours. For each step, an optimum mapping is obtained, taking into account closeness, curvature similarity and displacement on the contours. Evaluation with simulated contours showed that using interpolated contours improves the tracking. An increased number of intermediary contours slightly improves the results as well.