摘要
针对计算机辅助乳腺癌诊断中钙化点的提取具有较大难度的问题,计算了乳腺数字图像中每个像素点的最大梯度值。首先按照等梯度合并的原则将像素点划分为像素组,按照等分整个梯度区间的原则给定初始梯度聚类中心,模拟蚁群觅食行为中学习机制,以概率选择的方式进行像素点组的聚类得到新的聚类中心,然后再以模糊C均值法(FCM)对得到的聚类进行优化,从而提取出属于钙化点边缘的像素点。实验证明,通过选择适当的参数,用此方法提取乳腺钙化点边缘的效果良好。
Detection of mammary calcifications is challenged in computer-aided diagnosis. Maximum grade of each pixel of digital image of mammograph is calculated. Pixels are divided into different group and all pixels in each group have the same value of grade. Initial clustering centers are obtained through that grade area is divided into n equivalent parts. New clustering centers are obtained through simulating ant's search food behavior in which duster pixels group work acoording to probability selection principle. New clustering result are obtained through fuzzy C-means by using the above clustering centers as initialized centers. The experimental results demonstrate that the algorithm is effective to obtain edges of calcifications by choosing appropriate parameters.
出处
《光学技术》
EI
CAS
CSCD
北大核心
2008年第4期490-493,497,共5页
Optical Technique