摘要
计算机辅助诊断技术是提高诊断效率的有效手段,目前常使用肿瘤的形态和灰度纹理特征进行综合分析。而临床研究表明,肿瘤弹性也是判别其良恶性的重要指标。本文使用非刚性配准的方法,分析加压前后两幅灰阶超声肿瘤图像之间的差异,从而提取了形变总量和缩小放大比这两个反映肿瘤弹性的特征参数。随后的分类判决实验证明,这两个弹性参数对肿瘤良恶性具有较好的区分能力,联合使用形态特征后性能更优。
Computer aided diagnosis （CAD） is an effective technique to improve the efficiency of diagnosis, and today it usually performs a synthetic analysis based on the morphology and grayscale texture characteristics. However, clinical research demonstrated that the tumor elasticity was also a very important indicator to judge its benignity or malignance. In the current paper, a non-rigid registration method is used to analyze the difference between two grayscale ultrasonic images of the tumor before and after compression. Then, two characteristic parameters named total deformation and shrink-magnify ratio are extracted, reflecting the tumor elasticity. The classification experiment showed that the elasticity parameters had a good capability in determining whether the tumor was benign or malignant, and the performance could be improved by combining with morphology characteristics.
出处
《北京生物医学工程》
2008年第6期565-568,共4页
Beijing Biomedical Engineering
基金
安徽省教委自然科学基金重点研究项目(2006KJ097A)资助
关键词
乳腺肿瘤
超声图像
计算机辅助诊断
弹性特征
Breast tumor
ultrasound image
computer aided diagnosis
elasticity characteristics
