基于MCA的乳腺X线图像中肿块的自适应检测方法

An Adaptive Mass Detection Method on Mammography Based on MCA

针对肿块通常大小和形状各异,并且边缘模糊的特点,提出了一种基于形态学成分分析(MCA)和直方图自适应阈值搜索的肿块检测方法.首先通过引入MCA方法有效地抑制了血管和纤维对检测的影响,在此基础上设计了一种基于直方图的自适应阈值搜索策略,根据肿瘤的生长特性,通过自适应阈值和多灰度同心层方法,有效地检测乳腺X线图像中的病变区域.通过对真实乳腺X线图像的测试实验,其结果表明,所提出的方法能够检测出不同类型的肿块区域,并且假阳性区域的数量在可接受的范围内,能够有效地辅助医生进行诊断.

For capturing various shapes and blurry margins of tumors,a new mass detection method based on Morphological Component Analysis（MCA） and adaptive histogram threshold searching is proposed in this paper.Firstly,MCA method is introduced to restrain the influence of blood vessels and fibrous structures in mammograms.Then,an adaptive threshold searching method is designed according to the histograms of breast region.Finally,following the Gaussian-like growing feature of masses,the suspicious regions are effectively detected according to the adaptive thresholds and multi-intensity concentric layer methods.The experimental results on mammograms illustrate that the proposed method could effectively detect different types of masses with acceptable false positives,and could be a useful tool for assisting doctors.