基于双域自适应网络的岩矿样工业CT图像金属伪影校正算法研究

左顺吉; 冯鹏; 黄盼; 严笙豪; 何鹏; 魏彪

doi:10.15953/j.ctta.2022.041

基于双域自适应网络的岩矿样工业CT图像金属伪影校正算法研究

左顺吉^1,,
冯鹏^{1, 2},
黄盼¹,
严笙豪¹,
何鹏^{1, 2},
魏彪^{1, 2, ,}

1.
重庆大学光电技术及系统教育部重点实验室, 重庆400044
2.
重庆大学工业CT无损检测教育部工程研究中心, 重庆400044

基金项目: 科技部国家重点研发计划（重点锂、铍成矿带成矿规律与预测评价研究与综合（2019YFC0605203））；重庆市科委基础研究与前沿探索专项（自然科学基金）（面向小样本的管激发X射线荧光CT成像关键技术研究（cstc2020jcyj-msxmX0553））；重庆市科委技术创新与应用发展专项（轨道交通智慧化车站研究及应用（cstc2021jscx-gksbX0056））。

详细信息

作者简介:
左顺吉: 男，重庆大学光电工程学院硕士研究生，主要从事CT图像金属伪影校正研究，E-mail：1264119171@qq.com

魏彪: 男，博士，重庆大学教授，主要从事X射线、CT成像算法及工程应用等方面的研究，E-mail：weibiao@cqu.edu.cn

通讯作者:
魏彪*，男，博士，重庆大学教授，主要从事X射线、CT成像算法及工程应用等方面的研究，E-mail：weibiao@cqu.edu.cn

中图分类号: O 242；TP 391.41
计量
- 文章访问数: 592
- HTML全文浏览量: 130
- PDF下载量: 225
出版历程
- 收稿日期: 2022-03-10
- 修回日期: 2022-05-31
- 录用日期: 2022-06-06
- 网络出版日期: 2022-06-27
- 发布日期: 2022-11-02

Metal Artifact Reduction Algorithm for CT Images of Rock and Mineral Samples Based on Dual-domain Adaptive Network

ZUO Shunji^1,,
FENG Peng^{1, 2},
HUANG Pan¹,
YAN Shenghao¹,
HE Peng^{1, 2},
WEI Biao^{1, 2, ,}

1.
Key Laboratory of Optoelectronic Technology and System, Ministry of Education, Chongqing University, Chongqing 400044, China
2.
Engineering Research Center of Industrial CT Nondestructive Testing, Ministry of Education, Chongqing University, Chongqing 400044, China

摘要

摘要: 岩矿样中包含有大量高密度金属物质，致使其在工业CT图像上产生了金属伪影，严重影响岩矿样参数分析的准确性。为抑制岩矿样CT图像的金属伪影，本文提出一种基于双域自适应网络的岩矿样CT图像金属伪影校正算法（DDA-CNN-MAR），将含有金属伪影的CT图像分别通过投影域网络和图像域网络进行金属伪影的抑制，自适应融合双域处理结果，实现由含伪影图像到无伪影图像的端到端映射。该算法以残差编解码网络模型（RED-CNN）为基础，易于提取特征并恢复图像细节；双域结构可自适应调整投影域（伪影抑制）和图像域（细节修复）的权重，借以获得最优的校正结果。研究结果表明，较之于RED-CNN-MAR，经过DDA-CNN-MAR方法校正的图像，MSE减小2.570，而PSNR和SSIM则分别提高1.218 dB和0.018，有效提升岩矿样CT成像的图像质量。
- 工业CT /
- 深度学习 /
- 校正算法 /
- 金属岩矿样 /
- 金属伪影
Abstract: Rock and mineral samples contain a large amount of high-density metal substances, which often lead to metal artifacts in CT images and seriously affect the parameters analysis accuracy of rock and mineral samples. In order to improve the quality of CT images and suppress metal artifacts, in this paper we propose a dual-domain adaptive network-based metal artifact reduction algorithm for CT images of rock samples (DDA-CNN-MAR). The metal artifacts are suppressed by the projection domain network and the image domain network successively, and the dual domain processing results are adaptively fused to realize the end-to-end mapping from images with artifacts to artifact-free images. The algorithm is based on the residual encoder-decoder network model (RED-CNN), which is easy to extract features and restore image details. The dual-domain structure can adaptively adjust the weights of the projection domain (artifact suppression) and image domain (detail inpainting) to obtain optimal reduction results. The experimental results show that, compared with the RED-CNN metal artifact denoising method in the image domain, the MSE of the image corrected by the DDA-CNN-MAR method is reduced by 2.570, while the PSNR and SSIM are increased by 1.218 dB and 0.018 respectively, which effectively improves the CT image quality.
- industrial CT /
- deep learning /
- reduction algorithms /
- metal rock samples /
- metal artifacts

HTML全文

图 1 双域自适应网络结构原理图

Figure 1. Dual-domain adaptive convolutional neural network structure

下载: 全尺寸图片幻灯片

图 2 RED-CNN-MAR网络结构示意图

Figure 2. RED-CNN-MAR network structure

下载: 全尺寸图片幻灯片

图 3 一号样本的实验结果对比图

Figure 3. Comparison of experimental results of sample No.1

下载: 全尺寸图片幻灯片

图 4 为图3中（a）～（f）对应的区域局部放大图

Figure 4. The zoomed ROI area of Fig.3，respectively

下载: 全尺寸图片幻灯片

图 5 二号样本的实验结果对比图

Figure 5. Comparison of experimental results of sample No.2

下载: 全尺寸图片幻灯片

图 6 为图5中（a）～（f）对应的区域局部放大图

Figure 6. (a)-(f) are zoomed ROI area of Fig.5，respectively

下载: 全尺寸图片幻灯片

表 1 多种算法的定量评价结果一览表

Table 1 Quantitative evaluation results by multiple algorithms

指标	方法	1号样本	2号样本	平均值
SSIM	原始图像	0.877	0.861	0.854
	LI-NMAR	0.858	0.876	0.844
	RED-CNN-MAR	0.918	0.930	0.912
	$\widehat{{\boldsymbol{X}}}$	0.868	0.889	0.861
	本文算法	0.929	0.946	0.930
PSNR/dB	原始图像	22.308	21.698	22.985
	LI-NMAR	19.634	20.874	21.739
	RED-CNN-MAR	33.316	34.079	33.756
	$\widehat{{\boldsymbol{X}}}$	28.516	28.976	27.420
	本文算法	33.821	36.806	34.974
MSE	原始图像	6.498	7.442	6.811
	LI-NMAR	23.484	23.089	25.311
	RED-CNN-MAR	4.007	5.421	5.954
	$\widehat{{\boldsymbol{X}}}$	17.032	16.557	12.107
	本文算法	3.694	3.027	3.384

下载: 导出CSV

参考文献(17)

[1]	何鹏, 魏彪, 陈超, 等. 华南成矿省福建魁歧晶洞花岗岩样品孔隙结构的工业X-CT三维可视化研究[J]. 地质学报, 2014,88(4): 777−783. HE P, WEI B, CHEN C, et al. The three-dimensional visualization study on pore structure of Fujian Kuiqi geode granite sample in South China metallogenic belt based on industrial computed tomography[J]. Acta Geologica Sinica, 2014, 88(4): 777−783. (in Chinese).
[2]	TSAFNAT N, TSAFNAT G, JONES A S. Automated mineralogy using finite element analysis and X-ray microtomography[J]. Minerals Engineering, 2009, 22(2): 149−155. doi: 10.1016/j.mineng.2008.06.003
[3]	陈慧娟. CT系统中射束硬化校正算法研究[D]. 太原: 中北大学, 2010. CHEN H J. The algorithms study of beam-hardening correction in CT[D]. Taiyuan: North University of China, 2010. (in Chinese).
[4]	陈超, 魏彪, 梁婷, 等. 一种基于工业CT技术的岩芯样品孔隙度测量分析方法[J]. 物探与化探, 2013,37(3): 500−507. CHEN C, WEI B, LIANG T. et al. The application of industrial computation tomography (CT) to the analysis of core sample porosity[J]. Geophysical & Geochemical Exploration, 2013, 37(3): 500−507. (in Chinese).
[5]	肖文, 曾理. CT图像的金属伪影校正方法综述[J]. 中国体视学与图像分析, 2019,(1): 29−36. doi: 10.13505/j.1007-1482.2019.24.01.004 XIAO W, ZENG L. The review of metal artifact reduction for CT images[J]. Chinese Journal of Stereology and Image Analysis, 2019, (1): 29−36. (in Chinese). doi: 10.13505/j.1007-1482.2019.24.01.004
[6]	GJESTEBY L, MAN B D, JIN Y, et al. Metal artifact reduction in CT: Where are we after four decades?[J]. IEEE Access, 2016, 4: 5826−5849. doi: 10.1109/ACCESS.2016.2608621
[7]	CARON M, BOJANOWSKI P, JOULIN A, et al. Deep clustering for unsupervised learning of visual features[C]//Proceedings of the European Conference on Computer Vision (ECCV), 2018: 132-149.
[8]	REN S, HE K, GIRSHICK R, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[J]. Advances in Neural Information Processing Systems, 2015, 28: 91−99.
[9]	TANG Y, HUANG J, ZHANG F, et al. Deep residual networks with a fully connected reconstruction layer for single image super-resolution[J]. Neurocomputing, 2020, 405: 186−199. doi: 10.1016/j.neucom.2020.04.030
[10]	GJESTEBY L, YANG Q, XI Y, et al. Deep learning methods for CT image-domain metal artifact reduction[C]//Developments in X-ray Tomography XI. International Society for Optics and Photonics, 2017, 10391: 103910W.
[11]	马燕, 钟发生, 刘丰林. 基于条件生成式对抗网络的CT金属伪影校正研究[J]. 中国体视学与图像分析, 2021,26(2): 101−112. doi: 10.13505/j.1007-1482.2021.26.02.001 MA Y, ZHONG F S, LIU F L. CT metal artifact reduction based on conditional generative adversarial network[J]. Chinese Journal of Stereology and Image Analysis, 2021, 26(2): 101−112. (in Chinese). doi: 10.13505/j.1007-1482.2021.26.02.001
[12]	ZHANG Y, YU H. Convolutional neural network based metal artifact reduction in X-ray computed tomography[J]. IEEE Transactions on Medical Imaging, 2018, 37(6): 1370−1381. doi: 10.1109/TMI.2018.2823083
[13]	LIAO H, LIN W A, ZHOU S K, et al. ADN: Artifact disentanglement network for unsupervised metal artifact reduction[J]. IEEE Transactions on Medical Imaging, 2019, 39(3): 634−643.
[14]	马燕, 余海军, 钟发生, 等. 基于残差编解码网络的CT图像金属伪影校正[J]. 仪器仪表学报, 2020,41(8): 160−169. doi: 10.19650/j.cnki.cjsi.J2006503 MA Y, YU H J, ZHONG F S, et al. CT metal artifact reduction based on the residual encoder-decoder network[J]. Chinese Journal of Scientific Instrument, 2020, 41(8): 160−169. (in Chinese). doi: 10.19650/j.cnki.cjsi.J2006503
[15]	LIN W A, LIAO H, PENG C, et al. Dudonet: Dual domain network for ct metal artifact reduction[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019: 10512-10521.
[16]	CHEN H, ZHANG Y, KALRA M K, et al. Low-dose CT with a residual encoder-decoder convolutional neural network[J]. IEEE Transactions on Medical Imaging, 2017, 36(12): 2524−2535. doi: 10.1109/TMI.2017.2715284
[17]	KALENDER W A, HEBEL R, EBERSBERGER J. Reduction of CT artifacts caused by metallic implants[J]. Radiology, 1987, 164(2): 576−577. doi: 10.1148/radiology.164.2.3602406