Compared with the CV model, as a result of the combination of the gradient vector and neighborhood shape information, this semiautomatic segmentation method significantly improves the accuracy and robustness of AV segmentation, making it feasible for improved segmentation of aortic valves from US images that have fuzzy boundaries.

Image Guided Intervention for valvular heart disease is increasingly making progress in minimally invasive manner, where effective and accurate segmentation of aortic valve (AV) from echocardiography is fundamental to improve the intra operative location accuracy. This paper proposes a shape constraint Chan-Vese (CV) model for segmenting the AV from ultrasound (US) images. Considering the poor quality and speckle noise in AV US images, the problem of the overflow at the weak edge is solved by adding the shape constraint to the CV model. The predefined shape constructed from AV region is applied as an energy constraint to the energy function through a signed distance map, and the AV is detected from the US image by minimizing the energy function. A hundred AV segmentation results are analyzed in the experiment, where the evaluation parameters are 95.38 ± 2.7%, 1.4 ± 0.5 mm, 2.07 ± 1.3 mm in transthoracic AV and 97.21 ± 1.6%, 0.7 ± 0.15 mm, 1.04 ± 0.6 mm in transesophageal AV, which reveal that the shape constraint CV model can segment AV accurately, efficiently and robustly.