基于NSCT和CLAHE的乳腺钼靶X线图像微钙化点增强方法  

A novel microcalcification enhancement method for digital mammogram images based on NSCT and CLAHE

作　　者：谷宇[1,2] 吕晓琪[1,2] 吴凉 郝小静[2,3] 赵瑛[2] 喻大华[2] 张信雪 张文莉 黄显武[2] 任国印[2] 

机构地区：[1]上海大学计算机工程与科学学院,上海200444 [2]内蒙古科技大学信息工程学院内蒙古自治区模式识别与智能图像处理重点实验室,内蒙古包头014010 [3]内蒙古科技大学外国语学院,内蒙古包头014010

出　　处：《光学技术》2018年第1期6-12,共7页Optical Technique

基　　金：国家自然科学基金项目(61771266,81571753,81460279,81301281,61179019);内蒙古自治区自然科学基金项目(2015MS0604,2014MS0828);内蒙古自治区高等学校科学研究项目(NJZY145,NJZZ14161);包头市科技计划项目(2017C1002,2015C2006-14);内蒙古科技大学创新基金项目(2015QNGG03,2014QNGG08,2015QDL26,2014QDL045)

摘　　要：针对微钙化点容易漏检的问题,提出一种非下采样轮廓波变换结合对比度受限自适应直方图均衡的乳腺图像微钙化点增强新算法。对乳腺图像预处理,提取乳房区域并将胸肌区域去除;再对图像进行非下采样轮廓波变换提取多尺度、多方向的子带,对其中的多个高频子带采用高斯拉普拉斯算子检测边缘并增强;进一步采用对比度受限自适应直方图均衡算法,提高图像局部小区域的对比度,实现乳腺图像微钙化点增强算法。结果表明该方法是一种有效的乳腺钼靶图像微钙化点增强方法,为微钙化点检测和乳腺癌诊断提供支持。Because the microcalcifications are hard to be detected,a novel enhancement of microcalcifications in digital mammograms is proposed using non-subsampled contourlet transform combined with contrast limited adaptive histogram equalization.Mammogram image preprocessing is performed to extract breast region and remove pectoral region.The non-subsampled contourlet transform algorithm is adopted to decompose the mammogram into multidirectional and multiscale subbands.The edges are detected using Laplace of Gaussian operator and then are enhanced in the high frequency subbands.The contrast limited adaptive histogram equalization algorithm is further used to improve the contrast of the subimage,and mammogram enhancement algorithm is realized.The method can enhance the edges of the microcalcifications and improve the contrast between microcalcifications and mammary gland tissue.It is an effective method for enhancement of microcalcifications in digital mammograms,which can support the detection of microcalcifications and the diagnosis of breast cancer.

关 键 词：X射线光学 微钙化点 非下采样轮廓波变换 对比度受限自适应直方图均衡 图像增强 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]