[1]谭山湖,郭小燕,魏伟一.RDG-Net：基于双阶段解码器的结直肠息肉图像分割模型[J].中国医学物理学杂志,2025,42(1):52-58.[doi:DOI:10.3969/j.issn.1005-202X.2025.01.008]
　TAN Shanhu,GUO Xiaoyan,WEI Weiyi.RDG-Net: a colorectal polyp image segmentation model based on a dual-stage decoder[J].Chinese Journal of Medical Physics,2025,42(1):52-58.[doi:DOI:10.3969/j.issn.1005-202X.2025.01.008]
点击复制

RDG-Net：基于双阶段解码器的结直肠息肉图像分割模型()
分享到：

《中国医学物理学杂志》[ISSN:1005-202X/CN:44-1351/R]

卷:
42
期数:
2025年第1期
页码:
52-58
栏目:
医学影像物理
出版日期:
2025-01-19

文章信息/Info

Title:
RDG-Net: a colorectal polyp image segmentation model based on a dual-stage decoder
文章编号:
1005-202X（2025）01-0052-07
作者:
谭山湖1; 郭小燕1; 魏伟一2
1.甘肃农业大学信息科学技术学院， 甘肃 兰州730070； 2.西北师范大学计算机科学与工程学院， 甘肃 兰州730070
Author(s):
TAN Shanhu1;  GUO Xiaoyan1;  WEI Weiyi2
1. College of Information Science and Technology, Gansu Agricultural University, Lanzhou 730070, China 2. College of Computer Science and Engineering, Northwest Normal University, Lanzhou 730070, China
关键词:
图像分割; 结直肠息肉; 多尺度特征聚合; 并行卷积
Keywords:
Keywords: image segmentation colorectal polyp multi-scale feature aggregation parallel convolution
分类号:
R318
DOI:
DOI:10.3969/j.issn.1005-202X.2025.01.008
文献标志码:
A
摘要:
基于深度学习的息肉图像分割可以有效帮助医生评估癌前病变，本文针对结直肠息肉图像中息肉边界不清晰时分割效果不佳、对新样本范化能力不足的问题，提出一种基于双阶段解码器的结直肠息肉图像分割模型RDG-Net。该模型采用Res2Net-50作为编码器以提高图像分割精度。解码器分为两个阶段，第一阶段利用4层多尺度特征聚合模块整合不同阶段编码器提取的特征，第二阶段通过3层并行卷积融合模块增强解码器第一阶段输出的图像特征并解码至更高分辨率作为模型的最终输出结果。采用CVC-ClinicDB和Kvasir-SEG数据集的训练集数据进行模型训练，并采用CVC-ClinicDB与Kvasir-SEG数据集以及未参与训练的CVC300和ETIS-LaribPolypDB数据集分别对模型进行测试。测试结果显示，CVC-ClinicDB与Kvasir-SEG数据集上准确率、精确度、召回率、Dice系数、交并比和F2的平均值分别为98.41%、94.25%、92.62%、93.42%、87.69%、92.93%，CVC300和ETIS-LaribPolypDB数据集上各评价指标的平均结果分别为99.05%、87.79%、89.13%、88.39%、79.33%、88.82%。实验结果表明RDG-Net模型在结直肠息肉区域的分割任务中表现出色，在新数据集上表现出较好的泛化能力。
Abstract:
Abstract: Deep learning-based polyp image segmentation is helpful for assessing precancerous lesions. A colorectal polyp image segmentation model (RDG-Net) based on a dual-stage decoder is proposed to addresses the issues of poor segmentation performance due to unclear boundaries in colorectal polyp images and insufficient generalization ability for new samples. The model uses Res2Net-50 as the encoder to enhance the precision of image segmentation. The decoder has two stages: the first stage utilizes a 4-layer multi-scale feature aggregation module to integrate features extracted by the encoder at different stages, while the second stage enhances the image features output by the first stage of the decoder through a 3-layer parallel convolution fusion module and decodes them to a higher resolution as the models final output. The model is trained using the CVC-ClinicDB and Kvasir-SEG training datasets, and tested using the CVC-ClinicDB and Kvasir-SEG test datasets, as well as the CVC300 and ETIS-LaribPolypDB datasets that are not involved in the training. The test results show that the proposed method has an average accuracy, precision, recall rate, Dice similarity coefficient, intersection over union and F2 score of 98.41%, 94.25%, 92.62%, 93.42%, 87.69% and 92.93% on the CVC-ClinicDB and Kvasir-SEG datasets, while 99.05%, 87.79%, 89.13%, 88.39%, 79.33% and 88.82% on the CVC300 and ETIS-LaribPolypDB datasets, respectively, demonstrating that RDG-Net model performs well in colorectal polyp region segmentation and has a high generalization performance on new datasets.

相似文献/References:

[1]陈海斌,甄 鑫,周凌宏.基于先验的随机游走算法在医学图像分割中的应用[J].中国医学物理学杂志,2015,32(02):174.[doi:10.3969/j.issn.1005-202X.2015.02.005]
[2]许红玉,蔡坦坦,叶良凯,等.分水岭算法在CT图像分割中的应用[J].中国医学物理学杂志,2014,31(06):5272.[doi:10.3969/j.issn.1005-202X.2014.06.012]
[3]周露,张书旭,余辉,等.PET-CT图像配准的预处理研究[J].中国医学物理学杂志,2013,30(05):4392.[doi:10.3969/j.issn.1005-202X.2013.05.012]
[4]赖胜圣,刘虔铖,张刚平.基于模糊C均值自动随机游走算法在脑肿瘤分割中的应用[J].中国医学物理学杂志,2015,32(05):707.[doi:doi:10.3969/j.issn.1005-202X.2015.05.021]
[5]刘国才,胡泽田,朱苏雨,等.头颈部肿瘤PET与MRI融合放疗靶区自适应区域生长勾画[J].中国医学物理学杂志,2016,33(3):222.[doi:10.3969/j.issn.1005-202X.2016.03.002]
　[J].Chinese Journal of Medical Physics,2016,33(1):222.[doi:10.3969/j.issn.1005-202X.2016.03.002]
[6]贾高杰,邱崧,蔡茗名,等.三维点云重构和体显示在医学辅助诊断中的应用[J].中国医学物理学杂志,2016,33(6):593.[doi:10.3969/j.issn.1005-202X.2016.06.012]
　[J].Chinese Journal of Medical Physics,2016,33(1):593.[doi:10.3969/j.issn.1005-202X.2016.06.012]
[7]张泽凡,张东.基于交叉熵和GVF-Snake的子宫肌瘤高强度聚焦超声图像自动分割算法[J].中国医学物理学杂志,2016,33(8):776.[doi:10.3969/j.issn.1005-202X.2016.08.005]
　[J].Chinese Journal of Medical Physics,2016,33(1):776.[doi:10.3969/j.issn.1005-202X.2016.08.005]
[8]胡立伟,白凯,钟玉敏,等. 基于不同图像分割法构建3D打印右室双出口模型[J].中国医学物理学杂志,2016,33(12):1272.[doi:10.3969/j.issn.1005-202X.2016.12.020]
　[J].Chinese Journal of Medical Physics,2016,33(1):1272.[doi:10.3969/j.issn.1005-202X.2016.12.020]
[9]孟爽,王辉,谢蓄芬,等.超像素有偏观测模糊聚类的乳腺超声图像分割[J].中国医学物理学杂志,2017,34(7):693.[doi:10.3969/j.issn.1005-202X.2017.07.009]
　[J].Chinese Journal of Medical Physics,2017,34(1):693.[doi:10.3969/j.issn.1005-202X.2017.07.009]
[10]张玉芳,关天民,刘光孟,等. 基于CT数据的医学图像处理系统设计[J].中国医学物理学杂志,2019,36(9):1055.[doi:DOI:10.3969/j.issn.1005-202X.2019.09.012]
　ZHANG Yufang,GUAN Tianmin,LIU Guangmeng,et al. Design of medical image processing system based on CT data[J].Chinese Journal of Medical Physics,2019,36(1):1055.[doi:DOI:10.3969/j.issn.1005-202X.2019.09.012]

备注/Memo

备注/Memo:
【收稿日期】2024-09-18 【基金项目】国家自然科学基金（62363031） 【作者简介】谭山湖，硕士研究生，研究方向：深度学习、图像处理，E-mail: 964913737@qq.com 【通信作者】郭小燕，博士，教授，研究方向：深度学习、图像处理，E-mail: guoxy@gsau.edu.cn
更新日期/Last Update: 2025-01-19