一种基于改进的DDCM可形变模型的医学图像分割算法

A Medical Image Segmentation Algorithm Based on Improved DDCM Deformable Model

提出一种基于传统DDCM模型的具有自适应能力的可形变模型算法。利用局部归一化Snakes外力场的方法,提高模型的抗噪声能力,同时降低对初始位置的敏感性;构造图像边界信息图,用统计学方法获得Snakes的即时位置信息,解决传统模型的难于收敛到凹型边界、收敛条件不易控制以及移动步长的单一等问题。这种改进的可形变模型具有一定的自适应能力和较强的鲁棒性,而且效率高,在临床中有一定的实用价值。

An adaptive deformable model based on the conventional DDCM is proposed. The method of local normalization of the external force field was used to improve the antinoise ability of DDCM and reduce the sensibility to position of initial contour of Snakes to a certain extent. On the basis of a constructed edge map, instantaneous position information of Snakes was attained with the statistical method, while the information contributes to the solution to problems of difficulty in converging at concave edge and controlling the convergence condition and the problem of the step oneness. Compared with conventional DDCM, many experiments demonstrated that the improved algorithm was more ro- bust, adaptable and efficient. The improved algorithm is promising for clinic application systems.