一种基于相关矩阵的肺结节增强算法  

A correlation matrix-based enhancement filter for detecting pulmonary nodules

作　　者：苗芬[1] 吴龙海[1] 周荷琴[1] 

机构地区：[1]中国科学技术大学自动化系,合肥230027

出　　处：《北京生物医学工程》2008年第5期475-478,共4页Beijing Biomedical Engineering

摘　　要：针对肺结节的增强,提出了一种基于相关矩阵的多窗口图像增强算法,以提高肺结节检测的精确性。作为肺结节检测的预处理部分,本算法可以有效减小传统算法对噪声的敏感度,更好地增强肺结节并抑制血管。实验结果表明,本文所设计的增强算法对于增强合成图像以及实际的肺CT图像都有较好的效果,可以用于基于CT图像的辅助诊断系统的设计中。In this paper, a multi-window enhancement approach based on the eigenvalues of correlation matrix is proposed to improve the precision for puhnonary nodule detection. As a preprocessing step for nodule detection, this enhancement filter can suppress the sensitivity of conventional methods efficiently, and it is able to enhance nodules and suppress vessels in CT images. Experiment results show the filter works well for both synthesized and actual pulmonary CT images, and it will be applied to computer aided diagnosis （CAD） system.

关 键 词：图像增强 肺结节检测 HESSIAN矩阵 相关矩阵 CT图像 

分 类 号：R318.04[医药卫生—生物医学工程]