乳腺X线图像中致密区域肿块的检测算法  

Detection Algorithm on Mass Focus in Mammary Dense Region

作　　者：邢晓洁[1] 徐向民[1] 肖跃[1] 周丰乐[1] 

机构地区：[1]华南理工大学电子与信息学院,广州510640

出　　处：《微计算机信息》2009年第18期174-176,共3页Control & Automation

基　　金：基金申请人:徐向民;项目名称:医学视频数字化网络传输系统产业化研究;基金颁发部门:广东省科技厅;2007年省部产学研合作专项资金项目(2007B090400021)

摘　　要：提出一种在乳腺X线图像中致密区域肿块的检测算法。该算法根据乳腺致密区域肿块内核灰度一致性的特点,利用小波多尺度分析特性,在小波域的高频子带进行疑似肿块区域的搜索和定位;再借由分形维数对图像灰度表面粗糙度刻画能力对可疑病灶区判别筛选;最终针对致密区域肿块与正常腺体界限模糊的特性采用结合模糊理论的区域生长算法提取出保持关键诊断信息的肿块病灶区域。实验表明,提出的算法在乳腺致密区域肿块自动定位、肿块病灶区域边缘形态刻画等方面较传统算法有较大改进,能为后续的计算机辅助诊断决策判决提供有效全面的分析数据。A novel detecting algorithm aim at mass focus in mammary dense region of oriental female is proposed. This algorithm, based on the gray-scale coherence inside mass, conducts a search and location of possible focus region in the HF-subhands by the multi-scale analysis of wavelet, then carries out a selection for certain mass with the depiction of roughness of the gray-scale surface by Fractal dimension, and finally segments the mass focus region with blurry boundary using fuzzy region growing （FRG） algorithm to preserve key diagnosis information. With the improvement of automatic location of focus and its boundary depiction in the mammary dense region, this algorithm is capable to provide effective and comprehensive data for continued diagnosis decision analysis.

关 键 词：致密区域 肿块检测 小波分析 分形维数 模糊区域生长 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]