摘要
乳腺癌是一种严重危害人体健康的恶性肿瘤,准确的诊断对于预防和治疗乳腺癌至关重要。相比人工检测的方法,计算机辅助检测系统更为高效省时。研究基于迁移学习方法,在微调Alexnet模型的基础上采用基于核的支持向量机(SVM)作为分类器构建了Alexnet-SVM模型,使用该模型对BreakHis数据库中的乳腺肿瘤组织病理图像分类。为了进一步提高模型的分类准确率,使用GA、GWO、Grid三种算法对SVM的核参数进行了优化。结果表明,经过GA算法优化过后的SVM对BreakHis数据库中不同放大倍数(40,100,200,400)下的乳腺肿瘤组织病理图像平均分类准确率分别达到97.51%、97.65%、97.68%和97.12%。相比目前已有的深度神经网络模型,所提出的模型分类准确率更高。研究结果对于乳腺癌的早期诊断具有重要的临床应用价值。
Breast cancer is a kind of malignant tumor which seriously endangers human health.its accurate diagnosis is very important for the prevention and treatment of breast cancer.Compared to manual detection methods,computer-aided detec-tion systems are more efficient and time-saving.Based on the transfer learning method,this study uses kernel based support vector machine(SVM)as a classifier to construct the Alexnet SVM model on the basis of fine-tuning the Alexnet model,and uses this model to classify the pathological images of breast tumors in the BreakHis database.In order to further improve the classification accuracy of the model,this article uses GA,GWO,and Grid algorithms to optimize the kernel parameters of SVM.The results showed that the average classification accuracy of SVM optimized by GA algorithm for breast tumor tissue pathological images at different magnifications(40,100,200,400)in the BreakHis database reached 97.51%,97.65%,97.68%,and 97.12%,respectively.Compared to the existing deep neural network models,the proposed model has a higher classification accuracy.The research results have important clinical application value for the early diag-nosis of breast cancer.
作者
王振东
李悦
申炳俊
金丽虹
夏冰
WANG Zhendong;LI Yue;SHEN Bingjun;JIN Lihong;XIA Bing(School of Life Science and Technology,Changchun University of Science and Technology,Changchun 130022;Changchun Obstetrics-Gynecology Hospital,Changchun 130041)
出处
《长春理工大学学报（自然科学版）》
2023年第5期130-136,共7页
Journal of Changchun University of Science and Technology(Natural Science Edition)
基金
吉林省科技发展计划项目(20210204194YY)
吉林省教育厅科学技术研究项目(JJKH20220782KJ)。
关键词
乳腺癌
迁移学习
图像分类
支持向量机
优化算法
breast cancer
transfer learning
image classification
support vector machine
optimization algorithm
