基于重采样改进的多图谱分割算法  

Multi-atlas segmentation algorithm based on resampling

作　　者：夏瑞 马瑜 王文娜 罗宇卓 尚梦玉 XIA Rui, MA Yu, WANG Wen-na, LUO Yu-zhuo, SHANG Meng-yu(School of Physics and Electronic-Electrical Engineering, Ningxia University, Yinchuan 750021, Chin)

机构地区：[1]宁夏大学物理与电子电气工程学院,宁夏银川750021

出　　处：《计算机工程与设计》2018年第8期2587-2592,2659,共7页Computer Engineering and Design

基　　金：宁夏回族自治区自然科学基金项目(NZ1609);宁夏高校科学研究基金项目(NGY2016015);宁夏大学研究生创新研究基金项目(GIP2017009)

摘　　要：为在保证多图谱分割精确度的基础上,提高算法的时间效率,提出基于重采样改进的多图谱分割算法。利用重采样代替粗配准环节,在配准分割过程中,通过两次提取分割组织为中心的感兴趣区域,有效降低数据规模。在标签融合阶段利用K近邻搜素算法改进基于图像块的标签融合方法。实验结果表明,该算法实现了在较高的配准精度下,高效率得到分割结果。To improve the efficiency of the algorithm on the basis of ensuring the accuracy of multi-atlas segmentation,a multi-atlas segmentation method based on resampling was proposed.By using resampling instead of the coarse registration,the region of interest was extracted twice in the process of registration segmentation,so that the data size was reduced effectively.The label fusion method based on image patch was improved using K nearest neighbor search algorithm in the label fusion phase.Experimental results show that,the efficiency of the algorithm is improved while the segmentation accuracy is guaranteed.

关 键 词：多图谱配准 重采样 感兴趣区域 微分同胚 图像块 K近邻搜素 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]