[1]杨涛,黄淼,刘琮,等.基于迭代配准和感知损失的肺部伪CT合成深度学习算法[J].中国医学物理学杂志,2025,42(1):59-66.[doi:DOI:10.3969/j.issn.1005-202X.2025.01.009]
　YANG Tao,HUANG Miao,et al.Deep learning algorithm for lung CT synthesis based on iterative registration and perceptual loss[J].Chinese Journal of Medical Physics,2025,42(1):59-66.[doi:DOI:10.3969/j.issn.1005-202X.2025.01.009]
点击复制

基于迭代配准和感知损失的肺部伪CT合成深度学习算法()
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:
59-66
栏目:
医学影像物理
出版日期:
2025-01-19

文章信息/Info

Title:
Deep learning algorithm for lung CT synthesis based on iterative registration and perceptual loss
文章编号:
1005-202X（2025）01-0059-08
作者:
杨涛1; 2; 黄淼1; 刘琮3; 胡志华1; 陶莉莉1; 张淑平1
1.上海第二工业大学智能制造与控制工程学院， 上海 201209； 2.上海第二工业大学计算机与信息工程学院， 上海 201209； 3.上海商学院物联网工程系， 上海 201400
Author(s):
YANG Tao1;  2;  HUANG Miao1;  LIU Cong3;  HU Zhihua1;  TAO Lili1;  ZHANG Shuping1
1. School of Intelligent Manufacturing and Control Engineering, Shanghai Polytechnic University, Shanghai 201209, China 2. School of Computer and Information Engineering, Shanghai Polytechnic University, Shanghai 201209, China 3. School of Internet of Things Engineering, Shanghai Business School, Shanghai 201400, China
关键词:
锥形束CT; CycleGAN; 感知损失; Elastix; 图像合成
Keywords:
Keywords: cone beam computed tomography CycleGAN perceptual loss Elastix image synthesis
分类号:
R318
DOI:
DOI:10.3969/j.issn.1005-202X.2025.01.009
文献标志码:
A
摘要:
目的：使用深度学习算法通过学习肺部CT域图像特征，将锥形束CT（CBCT）合成高质量伪CT（sCT）图像。方法：本研究提出一种基于感知损失的循环生成式对抗网络模型（CycleGAN）和迭代配准的sCT生成算法，首先，结合感知损失和循环一致性损失来训练CycleGAN模型生成高质量的sCT图像，然后，利用Elastix配准工具对所生成的sCT图像和计划CT（pCT）图像进行配准，并用来迭代CycleGAN生成器模型。结果：在获取到的70例肺部肿瘤患者的pCT与CBCT数据上进行实验，从定量指标上看，利用本算法生成的sCT与pCT对比的结构相似度指标比CBCT与pCT对比的提升了11.9%，由0.825上升到0.923，均方绝对误差由110.97 HU降至78.62 HU，峰值信噪比由32.21 dB上升到34.74 dB，互信息由1.187上升到1.418。可视化评估中可见该算法大幅度消除了CBCT切片的散射伪影，突显骨质结构同时也修复了软组织结构。通过与当下流行的U-CycleGAN，R-CycleGAN和CUT模型对比，说明了本算法的有效性。结论：本文算法生成sCT图像能够有效减小CBCT与pCT间的剂量误差与结构误差，使其应用于精准的剂量计算，辅助医生的临床放疗诊断成为可能。
Abstract:
Abstract: Objective To synthesize high-quality synthetic CT (sCT) images from cone beam CT (CBCT) by learning lung CT domain image features with a deep learning algorithm. Methods A sCT generation algorithm which employs perceptual loss-based cyclic generative adversarial network model (CycleGAN) and iterative registration was presented. CycleGAN model was trained to generate high-quality sCT images by combining perceptual loss and cycle consistency loss and Elastix was used to register the generated sCT image and the planned CT (pCT) image, and iterate CycleGAN generator model. Results Experiments were conducted on the obtained pCT and CBCT data of 70 patients with lung tumors. From a quantitative perspective, the SSIM between sCT generated by the proposed algorithm and pCT was improved by 11.9% as compared with that between CBCT and pCT, increasing from 0.825 to 0.923 additionally, RMSE dropped from 110.97 HU to 78.62 HU, PSNR increased from 32.21 dB to 34.74 dB, and mutual information increased from 1.187 to 1.418. The visual evaluation revealed that the proposed algorithm greatly eliminated the scattering artifacts of CBCT slices, highlighted the bone structure, and repaired the soft tissue structure. The comparisons with U-CycleGAN, R-CycleGAN and CUT models confirmed the effectiveness of the proposed algorithm. Conclusion Using the proposed algorithm for sCT images generation can effectively reduce the dose error and structural error between CBCT and pCT, making it possible to apply the proposed algorithm to accurate dose calculations and assist doctors in clinical diagnosis.

相似文献/References:

[1]张俊俊,杨 振,柏朋刚,等.运动模体中靶区长度与锥形束CT扫描速度的相关性[J].中国医学物理学杂志,2015,32(02):259.[doi:10.3969/j.issn.1005-202X.2015.02.025]
[2]张一戈,邱小平,陈维军,等.基于图像引导技术对头颈部肿瘤临床摆位误差的剂量学影响[J].中国医学物理学杂志,2015,32(01):25.[doi:10.3969/j.issn.1005-202X.2015.01.007]
[3]李 毅,李文荣,苏 进,等.基于CBCT非小细胞肺癌外放边界研究[J].中国医学物理学杂志,2014,31(04):5006.[doi:10.3969/j.issn.1005-202X.2014.04.008]
[4]翁邓胡,王 建,尹中明,等.基于锥形束CT研究肺癌图像引导放疗的内靶区外放边界值[J].中国医学物理学杂志,2014,31(04):5012.[doi:10.3969/j.issn.1005-202X.2014.04.009]
[5]查元梓,吴晔明,陆冬青,等.应用锥形束CT研究儿童神经母细胞瘤放射治疗的摆位误差[J].中国医学物理学杂志,2014,31(05):5128.[doi:10.3969/j.issn.1005-202X.2014.05.007]
[6]沈奕晨,聂生东.锥形束CT与螺旋CT图像配准算法及其在放疗中的应用[J].中国医学物理学杂志,2014,31(05):5151.[doi:10.3969/j.issn.1005-202X.2014.05.013]
[7]林俊华,魏贤顶,孔旭东,等.kV级锥形束CT在图像引导放疗中的应用[J].中国医学物理学杂志,2014,31(05):5157.[doi:10.3969/j.issn.1005-202X.2014.05.014]
[8]周 琼,周剑良,张一戈,等.基于锥形束CT肺癌放射治疗两种体位固定技术摆位误差的研究[J].中国医学物理学杂志,2014,31(06):5258.[doi:10.3969/j.issn.1005-202X.2014.06.008]
[9]周永强,金献测,易金玲,等.锥形束CT的验收与QA探讨[J].中国医学物理学杂志,2013,30(01):3854.[doi:10.3969/j.issn.1005-202X.2013.01.005]
[10]刘浩,白靖平,李公平,等.锥形束CT图象引导放疗的应用性测试[J].中国医学物理学杂志,2013,30(01):3877.[doi:10.3969/j.issn.1005-202X.2013.01.010]

备注/Memo

备注/Memo:
【收稿日期】2024-06-26 【基金项目】上海市自然科学基金（20ZR1440300） 【作者简介】杨涛，硕士研究生，研究方向：医学影像处理，E-mail: 1079741347@qq.com 【通信作者】黄淼，副教授，研究方向：医学影像处理、深度学习、强化学习，E-mail: huangmiao@sspu.edu.cn
更新日期/Last Update: 2025-01-19