一种利用光子图像解决深度学习中子 CT
缺乏成对数据的方法

doi:10.3969/j.issn.1007-5461.2024.05.004

量子电子学报 ›› 2024, Vol. 41 ›› Issue (5): 738-751.doi: 10.3969/j.issn.1007-5461.2024.05.004

一种利用光子图像解决深度学习中子 CT 缺乏成对数据的方法

郭虎1,2, 陈帅2*, 杨明翰2, 张子恒2,3, 邵慧4, 汪建业2

( 1 安徽大学物质科学与信息技术研究院, 安徽合肥 230601; 2 中国科学院合肥物质科学研究院核能安全技术研究所, 安徽合肥 230031; 3 中国科学技术大学, 安徽合肥 230026; 4 安徽建筑大学, 安徽合肥 230022 )

收稿日期:2022-11-22 修回日期:2023-02-13 出版日期:2024-09-28 发布日期:2024-09-28
通讯作者: E-mail: shuai.chen@inest.cas.cn E-mail:E-mail: shuai.chen@inest.cas.cn
作者简介:郭虎 ( 1998 - ), 安徽亳州人, 研究生, 主要从事图像处理和计算机层析重建方面的研究。E-mail: gh694993146@mail.ustc.edu.cn
基金资助:
安徽省自然科学基金 (2108085QF285), 合肥市自然科学基金 (2021003), 安徽省古建筑智能感知与高维建模国际联合研究中心开放课题 (GJZZX2021KF04)

A method to solve lack of paired data in neutron computed tomography for deep learning by using photon images

GUO Hu1,2 , CHEN Shuai 2*, YANG Minghan2 , ZHANG Ziheng2,3 , SHAO Hui 4 , WANG Jianye2

( 1 Institutes of Physical Science and Information Technology, Anhui University, Hefei 230601, China; 2 Institute of Nuclear Energy Safety Technology, HFIPS, Chinese Academy of Sciences, Hefei 230031, China; 3 University of Science and Technology of China, Hefei 230026, China; 4 Anhui Jianzhu University, Hefei 230022, China )

Received:2022-11-22 Revised:2023-02-13 Published:2024-09-28 Online:2024-09-28

摘要/Abstract

摘要： 由于缺少高质量成对数据集, 深度学习在中子层析 (CT) 重建中的应用与发展受到了严重阻碍。虽然中子层析和光子层析成像原理都基于Radon变换, 但二者在粒子输运过程中的成像特征不同，所以光子层析图像训练的网络无法直接用于解决中子层析成像的重建问题。为此, 本文提出了一种可以解决光子和中子层析成像迁移过程中概率分布差异问题的无监督域适应网络。该方法通过引入最大均值差异计算以减小光子与中子层析图像特征之间的分布差异, 并通过小波变换与卷积神经网络相结合的方式增强重建的有效特征。与其他算法的对比验证表明, 该方法能够从低投影通量下的中子层析结果中重建出高质量的中子层析图像, 有效缓解了低投影通量下中子层析成像退化的问题。

关键词: 图像处理, 中子CT重建, 域适应迁移学习, 稀疏层析

Abstract: Due to the lack of high-quality paired datasets, the application and development of deep learning in neutron computed tomography (CT) reconstruction are severely hindered. Although the imaging principles of neutron CT and photon CT are both based on the Radon transform, the imaging characteristics of the two processes during particle transport are different, so the network trained for photon CT cannot be directly used to solve the reconstruction problem of neutron CT. Therefore, in this work, an unsupervised domain adaptive network is proposed that can solve the probability distribution difference problem in the migration process from photon tomography to neutron tomography. In the proposed method, the maximum mean difference is introduced to reduce the distribution difference between photon and neutron tomography image features, and furthermore, wavelet transform and convolution neural network are combined to enhance the effective features of reconstruction. The comparison experiments with other algorithms show that the proposed method can reconstruct highquality neutron tomography images from low-flux neutron tomography results, effectively alleviating the degradation of low-flux neutron tomography.

Key words: image processing, neutron computer tomography reconstruction, domain adaptive transfer learning, sparse tomography

中图分类号:

TP399

郭虎, 陈帅, 杨明翰, 张子恒, 邵慧, 汪建业. 一种利用光子图像解决深度学习中子 CT 缺乏成对数据的方法[J]. 量子电子学报, 2024, 41(5): 738-751.

GUO Hu, , CHEN Shuai , YANG Minghan , ZHANG Ziheng, , SHAO Hui , WANG Jianye. A method to solve lack of paired data in neutron computed tomography for deep learning by using photon images[J]. Chinese Journal of Quantum Electronics, 2024, 41(5): 738-751.

参考文献

[1] Hu Chen, Yi Zhang,Weihua Zhang, Peixi Liao, Ke Li, Jiliu Zhou, and GeWang. Low-dose ct via convolutional neural network. Biomedical Optics Express, 8, 2017.
[2] Eunhee Kang, Junhong Min, and Jong Chul Ye. A deep convolutional neural network using directional wavelets for low-dose x-ray ct reconstruction. Medical Physics, 44, 2017.
[3] Hamidreza Khodajou-Chokami, Seyed Abolfazl Hosseini, and Mohammad Reza Ay. A deep learning method for high-quality ultra-fast ct image reconstruction from sparsely sampled projections. Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 1029, 2022.
[4] Hanene Ben Yedder, Ben Cardoen, and Ghassan Hamarneh. Deep learning for biomedical image reconstruction: a survey. Artificial Intelligence Review, 54, 2021.
[5] Hai Miao Zhang and Bin Dong. A review on deep learning in medical image reconstruction. Journal ofthe Operations Research Society of China, 8, 2020.
[6] Peter Vontobel, Eberhard Lehmann, and William D. Carlson. Comparison of x-ray and neutron tomography investigations of geological materials. volume 52, 2005.
[7] John Banhart, András Borbély, Krzysztof Dzieciol, Francisco Garcia-Moreno, Ingo Manke, Nikolay Kardjilov, Anke Rita Kaysser-Pyzalla, Markus Strobl, and Wolfgang Treimer. X-ray and neutron imaging - complementary techniques for materials science and engineering. International Journal ofMaterials Research, 101, 2010.
[8] Salwa R. Soliman, Hala H. Zayed, Mazen M. Selim, H. Kasban, and T. Mongy. High quality reconstruction for neutron computerized tomography images. Alexandria Engineering Journal, 60, 2021.
[9] K Andersen. Neutron imaging and applications. Neutron Imaging and Applications, 2009.
[10] F. Pfeiffer, C. Grünzweig, O. Bunk, G. Frei, E. Lehmann, and C. David. Neutron phase imaging and tomography. Physical Review Letters, 96, 2006.
[11] J. M.C. Brown, U. Garbe, and D. Pelliccia. Statistical image reconstruction for high-throughput thermal neutron computed tomography. Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 942, 2019.
[12] Hyeung-Tae Kim, D. F. Tiana Razakamandimby R., Veronika Szilágyi, Zoltán Kis, László Szentmiklósi, Michal A. Glinicki, and Kyoungsoo Park. Reconstruction of concrete microstructure using complementarity of x-ray and neutron tomography. Cement and Concrete Research, 148:106540, 2021.
[13] Magnier L , Lauréline Lecarme, Alloin F , et al. Tomography Imaging of Lithium Electrodeposits Using Neutron, Synchrotron X-Ray, and Laboratory X-Ray Sources: A Comparison[J]. Frontiers in Energy Research, 2021, 9:657712.
[14] Venkatakrishnan S , Ziabari A , Hinkle J , et al. Convolutional neural network based non-iterative reconstruction for accelerating neutron tomography *[J]. Machine Learning: Science and Technology, 2021.
[15] David L. Donoho. Compressed sensing. IEEE Transactions on Information Theory, 52, 2006.
[16] Hengyong Yu and Ge Wang. Compressed sensing based interior tomography. Physics in Medicine and Biology, 54, 2009.
[17] Minji Lee, Yoseob Han, John Paul Ward, Michael Unser, and Jong Chul Ye. Interior tomography using 1d generalized total variation. part ii: Multiscale implementation. SIAM Journal on Imaging Sciences, 8, 2015.
[18] John PaulWard, Minji Lee, Jong ChulYe, and Michael Unser. Interior tomography using 1d generalized total variation. part i: Mathematical foundation. SIAM Journal on Imaging Sciences, 8, 2015.
[19] Jan Jakubek. Data processing and image reconstruction methods for pixel detectors. Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 576, 2007.
[20] Yoseob Han, Dufan Wu, Kyungsang Kim, and Quanzheng Li. Endto-end deep learning for interior tomography with low-dose x-ray ct. Physics in Medicine and Biology, 67(11):115001, may 2022.
[21] S. Zhang and E. Salari. Image denoising using a neural network based non-linear filter in wavelet domain. volume II, 2005.
[22] Viren Jain and H. Sebastian Seung. Natural image denoising with convolutional networks. 2009.
[23] PascalVincent, Hugo Larochelle, Isabelle Lajoie, Yoshua Bengio, and Pierre Antoine Manzagol. Stacked denoising autoencoders: Learning useful representations in a deep network with a local denoising criterion. Journal ofMachine Learning Research, 11, 2010.
[24] Mehdi Nasri and Hossein Nezamabadi-pour. Image denoising in the wavelet domain using a new adaptive thresholding function. Neurocomputing, 72, 2009.
[25] Xiao Jiao Mao, Chunhua Shen, and Yu Bin Yang. Image restoration using very deep convolutional encoder-decoder networks with symmetric skip connections. 2016.
[26] Jelmer M.Wolterink, Tim Leiner, Max A. Viergever, and Ivana Igum. Generative adversarial networks for noise reduction in low-dose ct. IEEE Transactions on Medical Imaging, 36, 2017.
[27] Ge Wang. A perspective on deep imaging. IEEE Access, 4, 2016.
[28] Golshan Mahmoudi and Hossein Ghadiri. Recent advances in x-ray ct image reconstruction techniques, 2019.
[29] J. M.C. Brown, U. Garbe, and D. Pelliccia. Statistical image reconstruction for high-throughput thermal neutron computed tomography. Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 942, 2019.
[30] Jeffrey Fessler. Statistical image reconstruction methods for transmission tomography, 2010.
[31] Liu C H, Xiong D F, Dong Y, et al. Optimization of temperature measurement of infrared radiation via theory of radiant heat transfer angle coefficient [J]. Chinese Journal ofQuantum Electronics, 2019, 36(4): 490-494.
刘纯红, 熊丹枫, 董勇, 等. 辐射换热角系数理论对红外辐射测温的优化[J]. 量子电子学报, 2019, 36(4): 490-494.
[32] Kaichao Liang, Hongkai Yang, and Yuxiang Xing. Comparison of projection domain, image domain, and comprehensive deep learning for sparse-view x-ray ct image reconstruction. arXiv preprint arXiv:1804.04289, 2018.
[33] Burkhard Schillinger and Francesco Grazzi. Artefacts in neutron ct - their effects and how to reduce some of them. volume 69, 2015.
[34] Yasushi Ikeda, Masanobu Yokoi, Masahiro Oda, Masayoshi Tamaki, and Gen Ichi Matsumoto. Correction of scattering neutron effects on neutron ct. Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 377, 1996.
[35] Muhammad Ghifary, W. Bastiaan Kleijn, and Mengjie Zhang. Domain adaptive neural networks for object recognition. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 8862, 2014.
[36] Muhammad Ghifary, W. Bastiaan Kleijn, Mengjie Zhang, David Balduzzi, and Wen Li. Deep reconstruction-classification networks for unsupervised domain adaptation. volume 9908 LNCS, 2016.
[37] Haris M , Shakhnarovich G , Ukita N . Deep Back-Projection Networks For Super-Resolution[J]. arXiv, 2018.
[38] Diederik P. Kingma and Jimmy Lei Ba. Adam: A method for stochastic optimization. 2015.
[39] Elbio Calzada, Florian Gruenauer, Martin Mühlbauer, Burkhard Schillinger, and Michael Schulz. New design for the antares-ii facility for neutron imaging at frm ii. Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 605, 2009.
[40] Cynthia H. McCollough, Adam C. Bartley, Rickey E. Carter, Baiyu Chen, Tammy A. Drees, Phillip Edwards, David R. Holmes, Alice E. Huang, Farhana Khan, Shuai Leng, Kyle L. McMillan, Gregory J. Michalak, Kristina M. Nunez, Lifeng Yu, and Joel G. Fletcher. Lowdose ct for the detection and classification of metastatic liver lesions: Results of the 2016 low dose ct grand challenge. Medical Physics, 44, 2017.
[41] Kyong Hwan Jin, Michael T. McCann, Emmanuel Froustey, and Michael Unser. Deep convolutional neural network for inverse problems in imaging. IEEE Transactions on Image Processing, 26, 2017.
[42] Hu Chen, Yi Zhang, Mannudeep K. Kalra, Feng Lin, Yang Chen, Peixi Liao, Jiliu Zhou, and Ge Wang. Low-dose ct with a residual encoder-decoder convolutional neural network. IEEE Transactions on Medical Imaging, 36, 2017.
[43] Kai Zhang, Wangmeng Zuo, Yunjin Chen, Deyu Meng, and Lei Zhang. Beyond a gaussian denoiser: Residual learning of deep cnn for image denoising. IEEE Transactions on Image Processing, 26, 2017.

一种利用光子图像解决深度学习中子 CT 缺乏成对数据的方法

A method to solve lack of paired data in neutron computed tomography for deep learning by using photon images

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	冉术坤, 席景科, 徐凯, 牛金龙. 一种量子彩色图像矢量中值滤波方案[J]. 量子电子学报, 2023, 40(6): 836-849.
[2]	朱尚超, 刘志飞, 魏战红, 臧一鸣, 孙文韬. 基于IBM Qiskit的QRMW模型及算法实现[J]. 量子电子学报, 2023, 40(5): 666-676.
[3]	董跃华, 张先光 . 二值形态学边缘检测算法的量子实现及仿真[J]. 量子电子学报, 2023, 40(5): 654-665.
[4]	於康杰, 方波, 李剑敏, 王震, 蔡晋辉, 邬佳璐, 何正龙 . 太赫兹成像三维块匹配自适应Canny边缘检测算法[J]. 量子电子学报, 2023, 40(4): 458-468.
[5]	葛宏义 , 王飞 , 蒋玉英 , 李丽 , 张元 , 贾柯柯 , . 基于宽度学习的太赫兹光谱图像小麦霉变识别研究[J]. 量子电子学报, 2023, 40(3): 360-368.
[6]	杜天宇, 高昆∗ , 吴朝. 光栅相衬 CT 中吸收信号环形伪影的去除方法[J]. 量子电子学报, 2023, 40(1): 40-47.
[7]	陈建明, , 曾祥津, , 钟丽云, , 邸江磊, ∗ , 秦玉文, ∗. 基于深度学习的图像配准方法研究进展综述[J]. 量子电子学报, 2022, 39(6): 899-926.
[8]	吴琼, 马雷∗. 一种基于 LoG 算子的量子图像边缘检测算法[J]. 量子电子学报, 2022, 39(5): 720-727.
[9]	杨一杉, 梁华为, 姚瑶, 陈帅∗. 金属遮挡条件下光子层析快速重构方法研究[J]. 量子电子学报, 2022, 39(4): 558-565.
[10]	张杨∗, 程正东, 朱斌. 基于改进 Faster R-CNN 的无人机小目标检测算法[J]. 量子电子学报, 2022, 39(3): 354-363.
[11]	毛仁祥, 常建华, 张树益李红旭, 张露瑶. 基于3D-MobileNetv2 的多目标实时跟踪框架[J]. 量子电子学报, 2022, 39(3): 364-372.
[12]	臧一鸣, 朱尚超, 魏战红∗, 刘志飞, 林小竹, 孙文韬. 一种量子图像伪彩色编码方法[J]. 量子电子学报, 2022, 39(3): 343-353.
[13]	洪富祥, 陈冲, 丘仲锋∗. 一种基于机器视觉的无人机同心圆靶精准降落方法[J]. 量子电子学报, 2021, 38(3): 307-315.
[14]	陈儒, 杨义勤, 秦凡凯, 孟昭晖, 苗昕扬, 赵昆, 詹洪磊∗. 样本内部缺陷的太赫兹光谱成像与算法优化[J]. 量子电子学报, 2021, 38(3): 265-271.
[15]	王昱皓, 唐泽恬, 钟岷哲, 王阳, 曾瑞敏, 朱登玮, 杨晨. 基于相位相关和纹理分类的SIFT 图像拼接算法[J]. 量子电子学报, 2020, 37(6): 650-658.