«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

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

卷:

41卷

期数:

2024年第12期

页码:

1517-1531

栏目:

医学影像物理

出版日期:

2024-12-17

文章信息/Info

Title:

Synthesis and classification of pulmonary nodules using two-stage-based generative adversarial network incorporating contextual transformer

文章编号:

1005-202X（2024）12-1517-15

作者:

尹智贤1; 4; 夏克文1; 张昭2; 贺紫平3

1.河北工业大学电子信息工程学院， 天津 300401； 2.天津中医药大学第一附属医院/国家中医针灸临床研究中心， 天津 300193； 3.长沙理工大学计算机与通信工程学院， 湖南 长沙 410114； 4.天津中德应用技术大学软件与通信学院， 天津 300350

Author(s):

YIN Zhixian1;  4;  XIA Kewen1;  ZHANG Zhao2;  HE Ziping3

1. School of Electronics and Information Engineering, Hebei University of Technology, Tianjin 300401, China 2. First Teaching Hospital of Tianjin University of Traditional Chinese Medicine/National Clinical Research Center for Chinese Medicine Acupuncture and Moxibustion, Tianjin 300193, China 3. School of Computer and Communication Engineering, Changsha University of Science and Technology, Changsha 410114, China 4. School of Software and Communication, Tianjin Sino-German University of Applied Sciences, Tianjin 300350, China

关键词:

肺结节生成; 上下文注意力; 生成对抗网络; 肺结节分类; CycleGAN

Keywords:

Keywords: pulmonary nodule synthesis contextual transformer generative adversarial network pulmonary nodule classification CycleGAN

分类号:

R318；TP391

DOI:

DOI:10.3969/j.issn.1005-202X.2024.12.009

文献标志码:

A

摘要:

提出一种融合上下文注意力的两段式生成对抗网络用于肺结节生成和分类。上下文注意力采用一种通道增强的多头上下文注意力机制，将通道注意力和多头上下文注意力结合，更好地处理特征图中的复杂语义关系，有效增强了模型的特征提取能力；两段式生成对抗网络框架用于实现肺结节在指定肺部区域的注入，该框架将生成任务分为两个阶段：第一阶段生成肺结节感兴趣区域图像，然后通过泊松融合模块与指定的肺实质进行融合，生成初始样本；第二阶段使用改进的CycleGAN模型对初始样本进行微调。同时，在判别器中引入跨层激励模块和辅助分类器实现对特征通道的再校正以及对肺结节的分类。在LIDC-IDRI数据集上进行实验验证，实验结果表明，所提方法在肺结节生成上的FID、IS和KID评分分别为115.153、2.619±0.095和0.062；在肺结节恶性度分类上准确率为70.23%，灵敏度、F1值和AUC分别为68.66%、68.92%和87.59%，表现出优于ADGAN等基于GAN的分类模型，以及VGG16等基准网络的性能。

Abstract:

Abstract: A two-stage-based generative adversarial network incorporating contextual transformer is proposed for synthesis and multiclass classification of pulmonary nodules. Contextual transformer adopts a channel-enhanced multi-head contextual transformer mechanism which combines channel attention and multi-head contextual transformer to better deal with the complex semantic relationship in the feature space, thereby effectively enhancing the feature extraction capability of the model. A two-stage-based generative adversarial network framework is used to achieve the injection of pulmonary nodules in the designated lung area, and divide the synthesis task into two stages. In the first stage, pulmonary nodule regions of interest images are generated and then fused with designated lung parenchyma through a Poisson blending module to generate preliminary samples in the second stage, an improved CycleGAN model is used to fine-tune the preliminary samples. Meanwhile, the skip layer excitation module and auxiliary classifier are introduced into the discriminator for realizing the re-correction of the feature channel and the classification of pulmonary nodules. Experiments on LIDC-IDRI dataset reveal that the proposed method has a FID, IS and KID of 115.153, 2.619±0.095 and 0.062 on pulmonary nodule synthesis, and achieves an accuracy, sensitivity, F1 value and AUC of 70.23%, 68.66%, 68.92% and 87.59% on pulmonary nodule malignancy classification, respectively, outperforming GAN-based classification models such as ADGAN, as well as benchmark networks such as VGG16.

备注/Memo

备注/Memo:

【收稿日期】2024-05-27 【基金项目】国家自然科学基金（42075129）；河北省自然科学基金（E2021202179）；河北省重点研发项目（21351803D）；河北省关键技术与产品研发专项（SJMYF2022Y06） 【作者简介】尹智贤，博士研究生，研究方向：医学图像处理、智能信息处理，E-mail: zhixian.yin@hotmail.com 【通信作者】夏克文，博士，教授，博士研究生导师，研究方向：智能信息处理、无线通信与智能天线，E-mail: kwxia@hebut.edu.cn

常用功能

更新日期/Last Update: 2024-12-20