摘要
研究了乳房肿瘤的分类问题.通过对500个已知肿瘤性质(良性或恶性)细胞核的显微图像的10个量化特征进行统计分析,利用模糊聚类法,建立了模糊聚类判别分类模型,给出了判别的阈值曲线,分析了最佳分界点,并得到了判别的阈值.进一步地,对未知乳房肿瘤性质的病例进行诊断.数值模拟表明:误判概率为6.2%,本文方法具有良好的应用前景.
In this paper, the problem of discriminating classification of breast tumor is studied. The statistical and quantization characteristic of 10 variables in the micro\|image of the 500 cell\|cores with known nature is analyzed. The method of fuzzy clustering is used. And the discriminating modeling of fuzzy clustering is established. During the course, the discriminating curve of threshold value is given. And based on the optimizing dividing point, the discriminating threshold value is obtained. Finally, the tumor of unknown nature is diagnosed and its feature is determined. The example indicates that the probability of mis\|discrimination is 6.2%.
出处
《装备指挥技术学院学报》
2002年第1期92-95,共4页
Journal of the Academy of Equipment Command & Technology
关键词
乳房癌
模糊聚类
误判概率
数学模型
诊断
breast cancer
fuzzy clustering
probability of mis\|discrimination
mathematical modeling
