High resolution cardiac shape registration using Ricci flow
Mingchen Gao, Rui Shi, Shaoting Zhang, Wei Zeng, Zhen Yu Qian, Xianfeng Daivd Gu, Dimitris Metaxas, Leon Axel · 2013
Current CT techniques are able to produce isotropic high resolution CT images (0.5mm). Recent research has revealed that the interior of the left ventricle has complex structures and topology, which has potentially valuable information. However, this makes the matching between models much more challenging. In this paper, we propose a novel method to match two models with non-trivial topology. 3D mesh models are flattened onto a 2D planar surfaces using discrete hyperbolic Ricci flow. Therefore, the 3D matching problem is converted to a much simpler 2D matching problem. We show the performance on the registration of high resolution left ventricle models.