【摘要】 在分析了目前肿瘤分类检测所采用的方法基础上,提出了一种基于特征空间分离变换(Eigenspace Separation Trans-form)结合支持向量机的乳癌识别新方法。用UCI数据库提供的569例乳腺肿瘤患者,乳腺肿块细胞核显微图像的30个量化特征样本集,进行了分类识别实验,结果表明采用新方法检测乳癌正确识别率达98.3%,优于传统的其他分类识别方法。更多还原

【Abstract】 This paper introduces the techniques and methods in breast tumor classification,presents a new method of breast cancer diagnosis based on eigenspace separation transform and support vetor machine.Classification experiments are conducted using sample set provided by UCI database with 569 cases of breast tumor cell each with 30 features.The results indicate that the accurate recognition rate of breast cancer is 98.3%.The new method is obviously superior to conventional methods.更多还原