Computer-aided diagnosis system of breast tumor ultrasound images
Computer-aided diagnosis system of breast tumor ultrasound images
复制标题
乳腺肿瘤超声图像计算机辅助诊断系统
DOI：
--
复制
发表时间：
2009
期刊：
影响因子：
--
通讯作者：
Wang, Qiang
中科院分区：
文献类型：
--
作者：
Li, Xiao-Feng;Shen, Yi;Wang, Qiang
提出了一种基于各向异性扩散和支持向量机的乳腺肿瘤超声图像计算机辅助诊断系统。关于肿瘤图像的纹理和几何形状共有26个特征，包括自相关（最后一个单词“autocorr”不完整，可能是“autocorrelation”自相关的缩写）
A computer-aided diagnosis system to process the breast tumor ultrasonic images was proposed based on the arisotropic diffusion and the support vector machine.Total 26 features about the textures and the geometry of the tumor images,including the autocorrelation coefficients,the mean square deviation coefficients,and the compact ratios,were extracted.These features not only reflect the essential differences between the malignant breast tumors and the benign ones,but also consider the characters of the medical ultrasonic images as the coherent images.Experimental results showed that the diagnosis accuracy of the system is 93.47%,its sensitivity is 94.3%,its specifity is 92.5%,its positive predictive value is 93.55% and negative predictive value is 93.4%,the area under the ROC curve is 0.9718.The overall performance of the proposed system is superior to that reported in the influential literature[1].