Please wait a minute...
浙江大学学报(工学版)
浙江大学学报(工学版)  2022, Vol. 56 Issue (9): 1789-1795    DOI: 10.3785/j.issn.1008-973X.2022.09.012
计算机与控制工程     
多尺度残差网络模型的开放式电阻抗成像算法
刘近贞1,2(),陈飞1,2,熊慧1,2
1. 天津工业大学 控制科学与工程学院，天津 300387
2. 天津工业大学 天津市电气装备智能控制重点实验室，天津 300387
Open electrical impedance imaging algorithm based on multi-scale residual network model
Jin-zhen LIU1,2(),Fei CHEN1,2,Hui XIONG1,2
1. School of Control Science and Engineering, Tiangong University, Tianjin 300387, China
2. Tianjin Key Laboratory of Intelligent Control of Electrical Equipment, Tiangong University, Tianjin 300387, China
 全文: PDF(1889 KB)   HTML
摘要：

针对开放式电阻抗成像（OEIT）的图像重建算法存在的成像精度低、对噪声敏感、重建图像伪影面积较大等问题，提出基于多尺度残差网络模型的OEIT算法. 该算法利用不同尺寸卷积核的残差块提取边界电压的多尺度特征；在完成特征拼接后，利用卷积实现深层信息融合，得到预测的电导率分布结果. 使用有限元法搭建OEIT正问题模型，构造“边界电压?电导率分布”数据集，将所提算法与其他算法在该数据集和实际模型实验中进行比较. 结果表明，所提算法使OEIT的重建精度、抗噪能力和定位目标准确性显著提高，并使检测目标的伪影面积缩小.
关键词： 开放式电阻抗成像（OEIT）;  图像重建;  深度学习;  残差网络;  多尺度特征    
Abstract:

An open electrical impedance tomography (OEIT) algorithm based on multi-scale residual neural network model was proposed, to improve the problems of OEIT image reconstruction algorithm, such as low imaging accuracy, sensitive to noise and large artifact area of reconstructed image. The algorithm used residual blocks with different sizes of convolution kernels to extract multi-scale features of boundary voltage. After the features were spliced, convolution was used to realize deep information fusion to obtain predicted conductivity distribution results. A model for the OEIT forward problem was built by the finite element method and a data set of "boundary voltage-conductivity distribution" was constructed. The proposed algorithm was compared with other algorithms in the data set and actual model experiments. Results show that the reconstruction accuracy, anti-noise ability and target location accuracy of OEIT are improved significantly by using the proposed algorithm, while the artifact area of the target is reduced.
Key words: open electrical impedance tomography (OEIT)    image reconstruction    deep learning    residual network    multi-scale feature
收稿日期: 2021-09-18 出版日期: 2022-09-28
CLC:  R 318.0  
基金资助: 天津市教委科研计划项目（2019KJ014）
作者简介: 刘近贞（1985—），女，副教授，博士，从事生物电磁信息检测与信号处理研究. orcid.org/0000-0003-0496-2859. E-mail： liujinzhen@tiangong.edu.cn
服务　　
把本文推荐给朋友
加入引用管理器
E-mail Alert
作者相关文章　　
刘近贞
陈飞
熊慧

引用本文:

刘近贞,陈飞,熊慧. 多尺度残差网络模型的开放式电阻抗成像算法[J]. 浙江大学学报(工学版), 2022, 56(9): 1789-1795.

Jin-zhen LIU,Fei CHEN,Hui XIONG. Open electrical impedance imaging algorithm based on multi-scale residual network model. Journal of ZheJiang University (Engineering Science), 2022, 56(9): 1789-1795.

链接本文:

https://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2022.09.012    或    https://www.zjujournals.com/eng/CN/Y2022/V56/I9/1789

图 1  开放式电阻抗成像测量方式示意图
图 2  原始残差网络的残差模块
图 3  多尺度残差模块
图 4  多尺度残差网络整体结构
组别 c1 c2 n
1 27, 25, 23 25 16
2 21, 19, 17 19 32
3 15, 13, 11 13 32
4 9, 7, 5 7 64
表 1  多尺度残差网络参数
图 5  3种深度学习开放式电阻抗成像算法的损失曲线
图 6  数据集部分样本模型图
图 7  4种开放式电阻抗成像算法的部分模型及其重建图像
算法 $\overline {{\rm{RIE}}} $ $\overline { {\rm{ICC} } }$ 算法 $\overline {{\rm{RIE}}} $ $\overline { {\rm{ICC} } }$
TV正则化 0.758 7 0.646 9 深度残差 0.250 7 0.860 8
1D-CNN 0.379 8 0.786 5 本研究 0.196 0 0.893 6
表 2  4种开放式电阻抗成像算法的评价指标对比
图 8  不同噪声下的部分模型及其重建图像
算法 SNR=80 dB SNR=50 dB SNR=30 dB
$\overline { {\rm{RIE} } } $ $\overline { {\rm{ICC} } } $ $\overline { {\rm{RIE} } } $ $\overline { {\rm{ICC} } } $ $\overline { {\rm{RIE} } } $ $\overline { {\rm{ICC} } } $
TV正则化 0.808 2 0.563 6 0.845 9 0.528 7 0.877 2 0.465 5
1D-CNN 0.378 0 0.787 9 0.378 3 0.787 6 0.401 5 0.770 9
深度残差 0.251 9 0.859 9 0.254 5 0.858 4 0.349 2 0.799 4
本研究 0.197 4 0.892 6 0.200 4 0.890 7 0.278 9 0.842 3
表 3  不同噪声的评价指标对比
图 9  测量系统硬件实物图
图 10  4种开放式电阻抗成像算法的图像重建结果
算法 $\overline { {\rm{RIE} } } $ $\overline { {\rm{ICC} } } $ 算法 $\overline { {\rm{RIE} } } $ $\overline { {\rm{ICC} } } $
TV正则化 0.812 9 0.543 3 深度残差 0.602 7 0.793 8
1D-CNN 0.664 1 0.737 4 本研究 0.556 1 0.814 3
表 4  实际数据的评价指标对比
1 JIANG D, WU Y, DEMOSTHENOUS A Hand gesture recognition using three-dimensional electrical impedance tomography[J]. IEEE Transactions on Circuits and Systems II: Express Briefs, 2020, 67 (9): 1554- 1558
doi: 10.1109/TCSII.2020.3006430
2 全卉, 马利庄 基于生物电阻抗的人体内脏脂肪预测方法[J]. 浙江大学学报: 工学版, 2011, 45 (2): 301- 305
QUAN Hui, MA Li-zhuang Visceral fat estimation method based on bioelectrical impedance technology[J]. Journal of Zhejiang University: Engineering Science, 2011, 45 (2): 301- 305
3 BAYFORD R, POLYDORIDES N Focus on recent advances in electrical impedance tomography[J]. Physiological Measurement, 2019, 40 (10): 100401
doi: 10.1088/1361-6579/ab42cd
4 LIU S, JIA J, ZHANG Y D, et al Image reconstruction in electrical impedance tomography based on structure-aware sparse Bayesian learning[J]. IEEE Transactions on Medical Imaging, 2018, 37 (9): 2090- 2102
doi: 10.1109/TMI.2018.2816739
5 YANG Z, YAN G Damage detection for composites using carbon nanotube film and electrical impedance tomography technology[J]. Advanced Experimental Mechanics, 2018, 3: 161- 166
6 ZHAO Z, CHANG M Y, FRERICHS I, et al Regional air trapping in acute exacerbation of obstructive lung diseases measured with electrical impedance tomography: a feasibility study[J]. Minerva Anestesiol, 2020, 86 (2): 172- 180
7 WITKOWSKA-WROBEL A, ARISTOVICH K, CRAWFORD A, et al Imaging of focal seizures with electrical impedance tomography and depth electrodes in real time[J]. Neuroimage, 2021, 234 (4): 117972
8 LIU D, KHAMBAMPATI A K, KIM S, et al Multi-phase flow monitoring with electrical impedance tomography using level set based method[J]. Nuclear Engineering and Design, 2015, 289: 108- 116
doi: 10.1016/j.nucengdes.2015.04.023
9 LIU D, KHAMBAMPATI A K, Du J A parametric level set method for electrical impedance tomography[J]. IEEE Transactions on Medical Imaging, 2018, 37 (2): 451- 460
doi: 10.1109/TMI.2017.2756078
10 ALBERTI G S, AMMARI H, JIN B, et al The linearized inverse problem in multifrequency electrical impedance tomography[J]. SIAM Journal on Imaging Sciences, 2016, 9 (4): 1525- 1551
doi: 10.1137/16M1061564
11 WANG J, HAN B, WAN W Elastic-net regularization for nonlinear electrical impedance tomography with a splitting approach[J]. Applicable Analysis, 2019, 98 (12): 2201- 2217
doi: 10.1080/00036811.2018.1451644
12 SANTOS T B R, NAKANISHI R M, KAIPIO J P, et al Introduction of sample based prior into the D-bar method through a schur complement property[J]. IEEE Transactions On Medical Imaging, 2020, 39 (12): 4085- 4093
doi: 10.1109/TMI.2020.3012428
13 QIN X, ZHANG Z, HUANG C, et al U2-Net: going deeper with nested U-structure for salient object detection [J]. Pattern Recognition, 2020, 106: 107404
doi: 10.1016/j.patcog.2020.107404
14 KO Y F, CHENG K S U-Net-based approach for automatic lung segmentation in electrical impedance tomography[J]. Physiological Measurement, 2021, 42 (2): 025002
doi: 10.1088/1361-6579/abe021
15 REN S, SUN K, TAN C, et al A two-stage deep learning method for robust shape reconstruction with electrical impedance tomography[J]. IEEE Transactions on Instrumentation and Measurement, 2020, 69 (7): 4887- 4897
doi: 10.1109/TIM.2019.2954722
16 JIN K H, MCCANN M T, FROUSTEY E, et al Deep convolutional neural network for inverse problems in imaging[J]. IEEE Transactions on Image Processing, 2017, 26 (9): 4509- 4522
doi: 10.1109/TIP.2017.2713099
17 CHEN Y, LI K, HAN Y Electrical resistance tomography with conditional generative adversarial networks[J]. Measurement Science and Technology, 2020, 31 (5): 055401
doi: 10.1088/1361-6501/ab62c4
18 HAMILTON S J, HAUPTMANN A Deep D-bar: real-time electrical impedance tomography imaging with deep neural networks[J]. IEEE Transactions on Medical Imaging, 2018, 37 (10): 2367- 2377
doi: 10.1109/TMI.2018.2828303
19 LI X, LU R, WANG Q, et al One-dimensional convolutional neural network (1D-CNN) image reconstruction for electrical impedance tomography[J]. Review of Scientific Instruments, 2020, 91 (12): 124704
doi: 10.1063/5.0025881
20 SUN B, YUE S, HAO Z, et al An improved Tikhonov regularization method for lung cancer monitoring using electrical impedance tomography[J]. IEEE Sensors Journal, 2019, 19 (8): 3049- 3057
doi: 10.1109/JSEN.2019.2892179
21 ZONG Z, WANG Y, WEI Z A review of algorithms and hardware implementations in electrical impedance tomography[J]. Progress In Electromagnetics Research, 2020, 169: 59- 71
doi: 10.2528/PIER20120401
22 LI F, TAN C, DONG F, et al V-Net deep imaging method for electrical resistance tomography[J]. IEEE Sensors Journal, 2020, 20 (12): 6460- 6469
doi: 10.1109/JSEN.2020.2973337
[1] 郝琨,王阔,王贝贝. 基于改进Mobilenet-YOLOv3的轻量级水下生物检测算法[J]. 浙江大学学报(工学版), 2022, 56(8): 1622-1632.
[2] 夏杰锋,唐武勤,杨强. 光伏航拍红外图像的热斑自动检测方法[J]. 浙江大学学报(工学版), 2022, 56(8): 1640-1647.
[3] 莫仁鹏,司小胜,李天梅,朱旭. 基于多尺度特征与注意力机制的轴承寿命预测[J]. 浙江大学学报(工学版), 2022, 56(7): 1447-1456.
[4] 赵永胜,李瑞祥,牛娜娜,赵志勇. 数字孪生驱动的机身形状控制方法[J]. 浙江大学学报(工学版), 2022, 56(7): 1457-1463.
[5] 何立,庞善民. 结合年龄监督和人脸先验的语音-人脸图像重建[J]. 浙江大学学报(工学版), 2022, 56(5): 1006-1016.
[6] 张雪芹,李天任. 基于Cycle-GAN和改进DPN网络的乳腺癌病理图像分类[J]. 浙江大学学报(工学版), 2022, 56(4): 727-735.
[7] 褚晶辉,史李栋,井佩光,吕卫. 适用于目标检测的上下文感知知识蒸馏网络[J]. 浙江大学学报(工学版), 2022, 56(3): 503-509.
[8] 程若然,赵晓丽,周浩军,叶翰辰. 基于深度学习的中文字体风格转换研究综述[J]. 浙江大学学报(工学版), 2022, 56(3): 510-519, 530.
[9] 陈彤,郭剑锋,韩心中,谢学立,席建祥. 基于生成对抗模型的可见光-红外图像匹配方法[J]. 浙江大学学报(工学版), 2022, 56(1): 63-74.
[10] 任松,朱倩雯,涂歆玥,邓超,王小书. 基于深度学习的公路隧道衬砌病害识别方法[J]. 浙江大学学报(工学版), 2022, 56(1): 92-99.
[11] 刘兴,余建波. 注意力卷积GRU自编码器及其在工业过程监控的应用[J]. 浙江大学学报(工学版), 2021, 55(9): 1643-1651.
[12] 陈雪云,黄小巧,谢丽. 基于多尺度条件生成对抗网络血细胞图像分类检测方法[J]. 浙江大学学报(工学版), 2021, 55(9): 1772-1781.
[13] 陈智超,焦海宁,杨杰,曾华福. 基于改进MobileNet v2的垃圾图像分类算法[J]. 浙江大学学报(工学版), 2021, 55(8): 1490-1499.
[14] 刘嘉诚,冀俊忠. 基于宽度学习系统的fMRI数据分类方法[J]. 浙江大学学报(工学版), 2021, 55(7): 1270-1278.
[15] 金立生,华强,郭柏苍,谢宪毅,闫福刚,武波涛. 基于优化DeepSort的前方车辆多目标跟踪[J]. 浙江大学学报(工学版), 2021, 55(6): 1056-1064.