Deep D-Bar: Real-Time Electrical Impedance Tomography Imaging With Deep Neural Networks

The mathematical problem for electrical impedance tomography (EIT) is a highly nonlinear ill-posed inverse problem requiring carefully designed reconstruction procedures to ensure reliable image generation. D-bar methods are based on a rigorous mathematical analysis and provide robust direct reconstructions by using a low-pass filtering of the associated nonlinear Fourier data. Similarly to low-pass filtering of linear Fourier data, only using low frequencies in the image recovery process results in blurred images lacking sharp features, such as clear organ boundaries. Convolutional neural networks provide a powerful framework for post-processing such convolved direct reconstructions. In this paper, we demonstrate that these CNN techniques lead to sharp and reliable reconstructions even for the highly nonlinear inverse problem of EIT. The network is trained on data sets of simulated examples and then applied to experimental data without the need to perform an additional transfer training. Results for absolute EIT images are presented using experimental EIT data from the ACT4 and KIT4 EIT systems.

[1]  Matteo Santacesaria,et al.  On an inverse problem for anisotropic conductivity in the plane , 2010, 1003.1880.

[2]  Jong Chul Ye,et al.  A deep convolutional neural network using directional wavelets for low‐dose X‐ray CT reconstruction , 2016, Medical physics.

[3]  C. Stocker,et al.  Ventilation and cardiac related impedance changes in children undergoing corrective open heart surgery , 2013, Physiological measurement.

[4]  G.J. Saulnier,et al.  An Electrical Impedance Spectroscopy System for Breast Cancer Detection , 2007, 2007 29th Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[5]  Giovanni Alessandrini,et al.  Stable determination of conductivity by boundary measurements , 1988 .

[6]  Ville Kolehmainen,et al.  Electrical Impedance Tomography Problem With Inaccurately Known Boundary and Contact Impedances , 2006, IEEE Transactions on Medical Imaging.

[7]  Caroline A Grant,et al.  Measurement of ventilation and cardiac related impedance changes with electrical impedance tomography , 2011, Critical care.

[8]  Ville Kolehmainen,et al.  Isotropic and anisotropic total variation regularization in electrical impedance tomography , 2017, Comput. Math. Appl..

[9]  David Isaacson,et al.  Reconstructions of chest phantoms by the D-bar method for electrical impedance tomography , 2004, IEEE Transactions on Medical Imaging.

[10]  Zhou Zhou,et al.  Comparison of total variation algorithms for electrical impedance tomography , 2015, Physiological measurement.

[11]  Nuutti Hyvönen,et al.  Approximating idealized boundary data of electric impedance tomography by electrode measurements , 2009 .

[12]  A. Nachman,et al.  Global uniqueness for a two-dimensional inverse boundary value problem , 1996 .

[13]  E. V. van Beek,et al.  Imaging of acute respiratory distress syndrome. , 2010, Expert opinion on medical diagnostics.

[14]  Andreas D. Waldmann,et al.  Position-dependent distribution of ventilation measured with electrical impedance tomography , 2015 .

[15]  Ning Liu.  ACT4: A high-precision, multi-frequency electrical impedance tomograph , 2007 .

[16]  E. Somersalo,et al.  Existence and uniqueness for electrode models for electric current computed tomography , 1992 .

[17]  Sébastien Martin,et al.  A Post-Processing Method for Three-Dimensional Electrical Impedance Tomography , 2017, Scientific Reports.

[18]  Dong Liu,et al.  A Parametric Level Set Method for Electrical Impedance Tomography , 2018, IEEE Transactions on Medical Imaging.

[19]  J. Borges,et al.  Real‐time ventilation and perfusion distributions by electrical impedance tomography during one‐lung ventilation with capnothorax , 2015, Acta anaesthesiologica Scandinavica.

[20]  J L Mueller,et al.  2D D-bar reconstructions of human chest and tank data using an improved approximation to the scattering transform. , 2010, Physiological measurement.

[21]  Matti Lassas,et al.  REGULARIZED D-BAR METHOD FOR THE INVERSE CONDUCTIVITY PROBLEM , 2009 .

[22]  Salvatore Grasso,et al.  Physiological Effects of the Open Lung Approach in Patients with Early, Mild, Diffuse Acute Respiratory Distress Syndrome: An Electrical Impedance Tomography Study , 2015, Anesthesiology.

[23]  Michael Unser,et al.  Deep Convolutional Neural Network for Inverse Problems in Imaging , 2016, IEEE Transactions on Image Processing.

[24]  Jennifer L. Mueller,et al.  Robust Computation of 2D EIT Absolute Images with D-bar Methods , 2017, ArXiv.

[25]  Neerav Dixit,et al.  Deep convolutional neural networks for accelerated dynamic magnetic resonance imaging , 2017 .

[26]  Andreas Hauptmann,et al.  Open 2D Electrical Impedance Tomography data archive , 2017, 1704.01178.

[27]  Andreas Hauptmann.  Approximation of full-boundary data from partial-boundary electrode measurements , 2017 .

[28]  Matti Lassas,et al.  A direct reconstruction method for anisotropic electrical impedance tomography , 2014, 1402.1117.

[29]  Melody Dodd,et al.  A Real-time D-bar Algorithm for 2-D Electrical Impedance Tomography Data. , 2014, Inverse problems and imaging.

[30]  Thomas Wiatowski,et al.  A Mathematical Theory of Deep Convolutional Neural Networks for Feature Extraction , 2015, IEEE Transactions on Information Theory.

[31]  Jérémi Dardé,et al.  Simultaneous Reconstruction of Outer Boundary Shape and Admittivity Distribution in Electrical Impedance Tomography , 2012, SIAM J. Imaging Sci..

[32]  G. Vainikko.  Fast Solvers of the Lippmann-Schwinger Equation , 2000 .

[33]  Jonas Adler,et al.  Solving ill-posed inverse problems using iterative deep neural networks , 2017, ArXiv.

[34]  Thomas Brox,et al.  U-Net: Convolutional Networks for Biomedical Image Segmentation , 2015, MICCAI.

[35]  Samuli Siltanen,et al.  Linear and Nonlinear Inverse Problems with Practical Applications , 2012, Computational science and engineering.

[36]  Jennifer L. Mueller,et al.  Effect of Domain Shape Modeling and Measurement Errors on the 2-D D-Bar Method for EIT , 2009, IEEE Transactions on Medical Imaging.

[37]  Andreas Hauptmann,et al.  Model-Based Learning for Accelerated, Limited-View 3-D Photoacoustic Tomography , 2017, IEEE Transactions on Medical Imaging.

[38]  Stephan Antholzer,et al.  Deep learning for photoacoustic tomography from sparse data , 2017, Inverse problems in science and engineering.

[39]  Jari P. Kaipio,et al.  Compensation of Modelling Errors Due to Unknown Domain Boundary in Electrical Impedance Tomography , 2011, IEEE Transactions on Medical Imaging.

[40]  Per Christian Hansen,et al.  Electrical impedance tomography: 3D reconstructions using scattering transforms , 2012 .