摘要
目的:为了提高计算机辅助诊断对肺结节良、恶性判断的精度,提出一种新的基于随机游走的肺结节分割方法。方法:首先,采用自适应中值滤波对图像进行平滑处理,消除困难肺结节内部灰度分布不均匀而导致的误分割;然后,将拉普拉斯零交叉点引入到随机游走算法权函数定义中,并根据图像中节点与种子点的距离来对图像进行边缘增强,消除弱边缘的干扰,获得外部特征清晰的肺结节分割结果。结果:与传统图像分割方法相比,该方法实现了三种困难肺结节的精确分割,对肺结节定量、定性分析提供更加准确的客观依据。结论:改进的随机游走图像分割可以有效地对困难肺结节进行精确分割。
Objective： To enhance the performance of computer aided diagnosis of the benign and malignant pulmonary nodules by an improved random walk based on pulmonary nodules segmentation method. Methods： Firstly, adaptive median filter algorithm was used to smooth the images and to solve the problem that the objective contour was easily influenced by the discontinuous distribution of intensity in Ground-Glass Opacity pulmonary nodules. Secondly, according to the distance between the node and the seed in the graph, Laplacian zero crossing was introduced into the weight function in random walker algorithm （RW） to enhance image edge. The interference of the weak edge could be eliminated, and a better segmentation result of pulmonary nodules could be obtained. Results： Compared with the traditional image segmentation methods, the proposed algorithm could achieve accurate segmentation of pulmonary nodules and provide more accurate and objective basis for the quantitative and qualitative analysis of pulmonary nodules. Conclusion： The improved random walker algorithm provides an effective method for accurate segmentation of the pulmonary nodules.
出处
《天津医科大学学报》
2014年第1期32-35,共4页
Journal of Tianjin Medical University
基金
国家自然科学基金资助项目(81000639)
中国博士后科学基金资助项目(20100470791
201104307)
天津医科大学校级基金资助项目(2009ky08)
关键词
计算机辅助诊断
随机游走
自适应中值滤波
边缘增强
图像分割
肺结节
computer aided diagnosis
random walker
, adaptive median filter
edge enhancement
image segmentation
pulmonary nodule
