Topological Distance-constrained Feature Descriptor Learning Model for Vessel Matching in Coronary Angiographies
Abstract — A vital requirement to establish the association between virtual and real objects in the field of virtual reality (VR) and augmented reality (AR) is feature matching. This technology is important for VR/AR display systems because it provides them with the ability to match the dynamic scene. Many methods for image matching, especially those based on deep learning techniques, have been proposed in the past few decades. However, vessel matching in coronary angiographies is an extremely difficult task because of the presence of vessel fracture, stenosis, artifacts, high background noise, and uneven vessel grayscale. The traditional matching methods performs poorly in such matching task. Methods In this study, a topological distance-constrained feature descriptor learning model is proposed. This model regards the topology of the vasculature as the connection relationship of the centerline. The topological distance combines the geodesic distance between the input patches and constrains the descriptor network by maximizing the feature difference between connected and unconnected patches to obtain more useful potential feature relationships. Results Matching patches of different sequences of angiographic images are generated for experiments. The proposed method has higher matching accuracy and stability than existing models. Conclusions The method generates a topological distance-constrained feature descriptor, which solves the problem of matching coronary angiographies.
