摘要
超声是目前诊断乳腺肿瘤的常用手段之一。针对超声中良恶性肿瘤纹理相似,区分度小等问题,论文提出了一个基于深度学习的乳腺超声良恶性自动诊断模型以辅助医生诊断。论文采用Densenet加强细节特征提取,注意力机制模拟临床诊断,迁移学习缓解数据依赖。实验结果表明,该模型可为年轻医生提供良好的辅助诊断,具有较好的可靠性和临床实用性。
Ultrasound is now one of the common means of diagnosing breast tumours.To address the problems of similar texture and low differentiation between benign and malignant tumours in ultrasound,this paper proposes a deep learning-based automatic diagnosis model for benign and malignant breast ultrasound to assist doctors in diagnosis.This paper uses Densenet to enhance detailed feature extraction,attention mechanisms to simulate clinical diagnosis,and transfer learning to alleviate data dependency.The experimental results show that the model can provide a good aid to diagnosis for young doctors,with good reliability and clinical utility.
作者
张宁
ZHANG Ning(College of Computer Science and Technology,China University of Petroleum(East China),Qingdao 266580)
出处
《计算机与数字工程》
2024年第2期427-431,共5页
Computer & Digital Engineering
关键词
乳腺肿瘤
超声图像
注意力机制
迁移学习
breast tumours
ultrasound images
attentional mechanisms
transfer learning