基于BBO优化BP神经网络的乳腺癌诊断

Breast Cancer Diagnosis Based on BP Neural Network Optimized by BBO

乳腺癌已成为世界上妇女发病率最高的癌症。医学研究发现,乳腺肿瘤病灶组织的细胞核显微图像与正常组织的细胞核显微图像不同,但是用一般的图像处理方法很难对其进行区分。因此,本文提出用生物地理学优化算法(BBO)优化BP神经网络对乳腺癌进行诊断,将乳腺肿瘤病灶组织的细胞核显微图像的10个量化特征作为网络的输入,良性乳腺肿瘤和恶性乳腺肿瘤作为网络的输出。用训练集数据对设计的BBOBP神经网络进行训练,然后对测试集数据进行测试并对测试结果进行分析。结果表明BBOBP有很好的分类性能,能对乳腺癌进行有效的诊断,且误诊率较低。

Breast cancer has become the most common cancer among women in the world. Medical studies have found that the nuclear nficrograph of breast tumor lesion tissue is different from that of nomlal tissue, but it is difficult to distinguish it by general image processing method. Therefore, this article puts forvard a method by using biogeography optimization algorithm （BBO） to optimize the BP neural network for the diagnosis of breast cancer, 10 quantitative characteristics of the nucleus microscopic images input and benign breast tumor ral network is trained with the and malignant breast tumor as the training set data, and then the test of the b^ast tumor lesion group takes as network output of the network. The designed BBOBP neu- set data is tested and the test ~sults are analyzed. The results show that BBOBP has good classification perfommnce that can effectively diagnose breast cancer, and the nfisdiagnosis rate is low.