Aortic Valve Ultrasound Image Segmentation Based On Improved Active Contour Model

Effective and accurate segmentation of Aortic Valve (AV) from sequenced ultrasound(US) images remains a technical challenge because of intrinsic factors of ultrasound imagesimpacting the quality and the continuous changes of shape and position of segmented objects.In this paper, a novel shape constraint gradient Chan-Vese (GCV) model is proposed forsegmenting AV from time serial echocardiography.The GCV model is derived from incorporating the energy of gradient vector flow intoChan-Vese (CV) model framework, where the gradient vector energy term is introduced bycalculating the deviation angle between the inward normal force of the evolution contour andgradient vector force. The flow force enlarges the capture range and enhances the blurredboundary of objects. While circle-like contour, constructed from AV structure region asconstraint shape, is added as energy items to the GCV model through the shape comparisonfunction, the shape-constrained energy can enhance the image constraint force for connectingseparate gaps of the object edge effectively as well as driving the evolution contour toapproach ideal object quickly. Considering the slight movement of AV in adjacent frames, theinitial constraint shape is defined by users and other constraint shapes are derived from thesegmentation results of adjacent sequence frames after a morphological filtering. Minimizingthe proposed energy function can achieve AV detection results from US images.To evaluate the performance of the proposed method, five assessment parameters areutilized to compare it with radiologist’s manual delineations (gold standards). The proposedsemiautomatic segmentation method significantly improved accuracy and robustnessbeneficial from the combination of the gradient vector flow and neighborhood shapeinformation, where it is feasible for better segmenting AV boundaries from US images withfuzzy boundaries.