中文摘要：

冠状动脉分割与狭窄分级、斑块检测等密切相关,是血管病变研究中的重要步骤。针对血管灰度不均和对比度低等问题,提出了一种基于活动窄带和符号压力函数水平集的CT血管造影冠状动脉分割方法。首先对初始轮廓做形态学膨胀和腐蚀运算,以构建活动窄带限定轮廓曲线的演化区域;其次在活动窄带区域内构造局部符号压力函数,用水平集算法使初始轮廓收敛至准确轮廓;最后利用形态学闭运算平滑曲线。通过利用活动窄带将图像区域局部化,降低了计算复杂度,克服了灰度不均匀性,促进轮廓曲线演化到细小的血管末梢和狭窄区域。实验结果表明,与传统的分割方法相比,能够更加有效准确地分割出冠状动脉,为血管病变的研究提供支持。

英文摘要：

As an important step in vascular disease study, segmentation of the coronary artery images is closely associated with stenosis grading and plaque detection. Herein we propose a new method for segmentation of CT angiographic（CTA） coronary artery images by combining the active narrow band and sign pressure function（SPF） level set to solve the problems of gray scale heterogeneity and low contrast of the vessels. Morphological dilation and erosion operations were applied to the initial contour to establish an active narrow band to limit the contour evolution. A local SPF was then constructed in the active narrow band, and a level set algorithm was used to facilitate the convergence of the initial contour into the exact contour. Finally, a morphological closing operation was utilized to smooth the curve. With the use of active narrow band to localize the image region, the computational complexity and the gray scale heterogeneity of the images were reduced to promote the evolution of the contour curve to the tiny peripheral vessels and narrow areas. The experimental results showed that compared with the traditional segmentation method, the proposed method achieved more accurate segmentation of the coronary artery images to facilitate the diagnosis of the vascular lesions.