摘要
根据近年来国内外计算机辅助检测(computer-aided detection,CAD)技术在CT图像肺结节检测中的研究进展情况,本文对比分析了目前检测流程中6个阶段(图像采集、预处理、肺实质分割、感兴趣区域提取、特征提取与优化、检测分析与降低假阳性率)各自所运用的研究方法及问题,并提出肺结节检测准确率的提高,依赖于各步骤算法的优化和大样本标准病例数据库的建立,需要在研究针对单一类型结节分类算法的基础上,设计通用的结节分类算法。
Based on the research progress of computer aided detection( CAD) technology forpulmonary nodules in CT scans in recent years,we analyzed the research methods and issues in the six detectingstages: image acquisition, pre-processing, lung parenchyma segmentation, regions of interest extraction,characteristics extraction and optimization,nodule detection and false positive reduction. Finally,we proposedthat optimizing the algorithms in each stage and establishing clinical database with large samples can improvethe detection accuracy. Meanwhile, design of universal nodule detection algorithm is an important researchdirection.
出处
《北京生物医学工程》
2016年第1期81-86,共6页
Beijing Biomedical Engineering
基金
国家自然科学基金(60972122)
上海市自然科学基金(14ZR1427900)
上海市研究生创新基金(JWCXSL1402)资助
关键词
肺结节
计算机辅助检测
CT图像
pulmonary nodule
computer-aided detection
CT image