[1]孙芳芳△,张 玲,梁乐平.基于迁移学习的超声图像甲状腺结节定位方法[J].生物医学工程研究,2020,04:347-352.
SUN Fangfang,ZHANG Ling,LIANG Leping.Thyroid nodules localization method for ultrasound images[J].Journal of Biomedical Engineering Research,2020,04:347-352.
点击复制
基于迁移学习的超声图像甲状腺结节定位方法(PDF)
《生物医学工程研究》[ISSN:1006-6977/CN:61-1281/TN]
- 期数:
-
2020年04期
- 页码:
-
347-352
- 栏目:
-
- 出版日期:
-
2020-12-25
文章信息/Info
- Title:
-
Thyroid nodules localization method for ultrasound images
- 文章编号:
-
1672-6278 (2020)04-0347-06
- 作者:
-
孙芳芳△; 张 玲; 梁乐平
-
空军军医大学第二附属医院,西安710038
- Author(s):
-
SUN Fangfang; ZHANG Ling; LIANG Leping
-
The Second Affiliated Hospital, Air Force Military Medical University, Xi′an 710038, China
-
- 关键词:
-
甲状腺结节; 超声图像; 卷积神经网络; 迁移学习; 目标定位
- Keywords:
-
Thyroid nodules; Ultrasound images; Convolutional neural network; Transfer learning; Objection localization
- 分类号:
-
R318
- DOI:
-
10.19529/j.cnki.1672-6278.2020.04.05
- 文献标识码:
-
A
- 摘要:
-
本研究提出一种基于迁移学习的甲状腺结节定位方法,利用深度卷积神经网络提取结节超声图像的特征,进而采用包围盒回归的方式定位甲状腺结节。分别分析了基于Xception、VGG-19和Resnet50三种预训练模型的结节定位方法。结果表明,基于Resnet50模型的神经网络结构在小样本量条件下,具有较高的定位准确率,有一定的临床应用价值。
- Abstract:
-
We proposed a new method of thyroid nodule localization based on transfer learning. The feature of ultrasound image was extracted by deep convolution neural network, and then the thyroid nodule was located by bounding box regression. The performance of nodules localization methods based on Xception, VGG-19 and Resnet50 models were analyzed and compared. The experimental results show that the neural network structure based on Resnet50 model has high positioning accuracy under the condition of small sample size, and has clinical application value.
备注/Memo
- 备注/Memo:
-
(收稿日期:2020-03-15)△通信作者Email:pepper24@163.com
更新日期/Last Update:
2021-02-07