A Rotational Invariant Neural Network for Electrical Impedance Tomography Imaging without Reference Voltage: RF-REIM-NET
<i>Background</i>: Electrical Impedance Tomography (EIT) is a radiation-free technique for image reconstruction. However, as the inverse problem of EIT is non-linear and ill-posed, the reconstruction of sharp conductivity images poses a major problem. With the emergence of artificial neu...
Main Authors: | , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2022-03-01
|
Series: | Diagnostics |
Subjects: | |
Online Access: | https://www.mdpi.com/2075-4418/12/4/777 |
_version_ | 1797436104826159104 |
---|---|
author | Jöran Rixen Benedikt Eliasson Benjamin Hentze Thomas Muders Christian Putensen Steffen Leonhardt Chuong Ngo |
author_facet | Jöran Rixen Benedikt Eliasson Benjamin Hentze Thomas Muders Christian Putensen Steffen Leonhardt Chuong Ngo |
author_sort | Jöran Rixen |
collection | DOAJ |
description | <i>Background</i>: Electrical Impedance Tomography (EIT) is a radiation-free technique for image reconstruction. However, as the inverse problem of EIT is non-linear and ill-posed, the reconstruction of sharp conductivity images poses a major problem. With the emergence of artificial neural networks (ANN), their application in EIT has recently gained interest. <i>Methodology</i>: We propose an ANN that can solve the inverse problem without the presence of a reference voltage. At the end of the ANN, we reused the dense layers multiple times, considering that the EIT exhibits rotational symmetries in a circular domain. To avoid bias in training data, the conductivity range used in the simulations was greater than expected in measurements. We also propose a new method that creates new data samples from existing training data. <i>Results</i>: We show that our ANN is more robust with respect to noise compared with the analytical Gauss–Newton approach. The reconstruction results for EIT phantom tank measurements are also clearer, as ringing artefacts are less pronounced. To evaluate the performance of the ANN under real-world conditions, we perform reconstructions on an experimental pig study with computed tomography for comparison. <i>Conclusions</i>: Our proposed ANN can reconstruct EIT images without the need of a reference voltage. |
first_indexed | 2024-03-09T10:57:51Z |
format | Article |
id | doaj.art-0c533dfce9df4532a916b70dcdcd83ff |
institution | Directory Open Access Journal |
issn | 2075-4418 |
language | English |
last_indexed | 2024-03-09T10:57:51Z |
publishDate | 2022-03-01 |
publisher | MDPI AG |
record_format | Article |
series | Diagnostics |
spelling | doaj.art-0c533dfce9df4532a916b70dcdcd83ff2023-12-01T01:29:02ZengMDPI AGDiagnostics2075-44182022-03-0112477710.3390/diagnostics12040777A Rotational Invariant Neural Network for Electrical Impedance Tomography Imaging without Reference Voltage: RF-REIM-NETJöran Rixen0Benedikt Eliasson1Benjamin Hentze2Thomas Muders3Christian Putensen4Steffen Leonhardt5Chuong Ngo6Helmholtz Institute for Biomedical Engineering, RWTH Aachen University, 52074 Aachen, GermanyHelmholtz Institute for Biomedical Engineering, RWTH Aachen University, 52074 Aachen, GermanyHelmholtz Institute for Biomedical Engineering, RWTH Aachen University, 52074 Aachen, GermanyDepartment of Anaesthesiology and Intensive Care Medicine, University of Bonn, Venusberg-Campus 1, 53127 Bonn, GermanyDepartment of Anaesthesiology and Intensive Care Medicine, University of Bonn, Venusberg-Campus 1, 53127 Bonn, GermanyHelmholtz Institute for Biomedical Engineering, RWTH Aachen University, 52074 Aachen, GermanyHelmholtz Institute for Biomedical Engineering, RWTH Aachen University, 52074 Aachen, Germany<i>Background</i>: Electrical Impedance Tomography (EIT) is a radiation-free technique for image reconstruction. However, as the inverse problem of EIT is non-linear and ill-posed, the reconstruction of sharp conductivity images poses a major problem. With the emergence of artificial neural networks (ANN), their application in EIT has recently gained interest. <i>Methodology</i>: We propose an ANN that can solve the inverse problem without the presence of a reference voltage. At the end of the ANN, we reused the dense layers multiple times, considering that the EIT exhibits rotational symmetries in a circular domain. To avoid bias in training data, the conductivity range used in the simulations was greater than expected in measurements. We also propose a new method that creates new data samples from existing training data. <i>Results</i>: We show that our ANN is more robust with respect to noise compared with the analytical Gauss–Newton approach. The reconstruction results for EIT phantom tank measurements are also clearer, as ringing artefacts are less pronounced. To evaluate the performance of the ANN under real-world conditions, we perform reconstructions on an experimental pig study with computed tomography for comparison. <i>Conclusions</i>: Our proposed ANN can reconstruct EIT images without the need of a reference voltage.https://www.mdpi.com/2075-4418/12/4/777artificial intelligencedeep learningElectrical Impedance Tomographylung imagingcardiopulmonary monitoring |
spellingShingle | Jöran Rixen Benedikt Eliasson Benjamin Hentze Thomas Muders Christian Putensen Steffen Leonhardt Chuong Ngo A Rotational Invariant Neural Network for Electrical Impedance Tomography Imaging without Reference Voltage: RF-REIM-NET Diagnostics artificial intelligence deep learning Electrical Impedance Tomography lung imaging cardiopulmonary monitoring |
title | A Rotational Invariant Neural Network for Electrical Impedance Tomography Imaging without Reference Voltage: RF-REIM-NET |
title_full | A Rotational Invariant Neural Network for Electrical Impedance Tomography Imaging without Reference Voltage: RF-REIM-NET |
title_fullStr | A Rotational Invariant Neural Network for Electrical Impedance Tomography Imaging without Reference Voltage: RF-REIM-NET |
title_full_unstemmed | A Rotational Invariant Neural Network for Electrical Impedance Tomography Imaging without Reference Voltage: RF-REIM-NET |
title_short | A Rotational Invariant Neural Network for Electrical Impedance Tomography Imaging without Reference Voltage: RF-REIM-NET |
title_sort | rotational invariant neural network for electrical impedance tomography imaging without reference voltage rf reim net |
topic | artificial intelligence deep learning Electrical Impedance Tomography lung imaging cardiopulmonary monitoring |
url | https://www.mdpi.com/2075-4418/12/4/777 |
work_keys_str_mv | AT joranrixen arotationalinvariantneuralnetworkforelectricalimpedancetomographyimagingwithoutreferencevoltagerfreimnet AT benedikteliasson arotationalinvariantneuralnetworkforelectricalimpedancetomographyimagingwithoutreferencevoltagerfreimnet AT benjaminhentze arotationalinvariantneuralnetworkforelectricalimpedancetomographyimagingwithoutreferencevoltagerfreimnet AT thomasmuders arotationalinvariantneuralnetworkforelectricalimpedancetomographyimagingwithoutreferencevoltagerfreimnet AT christianputensen arotationalinvariantneuralnetworkforelectricalimpedancetomographyimagingwithoutreferencevoltagerfreimnet AT steffenleonhardt arotationalinvariantneuralnetworkforelectricalimpedancetomographyimagingwithoutreferencevoltagerfreimnet AT chuongngo arotationalinvariantneuralnetworkforelectricalimpedancetomographyimagingwithoutreferencevoltagerfreimnet AT joranrixen rotationalinvariantneuralnetworkforelectricalimpedancetomographyimagingwithoutreferencevoltagerfreimnet AT benedikteliasson rotationalinvariantneuralnetworkforelectricalimpedancetomographyimagingwithoutreferencevoltagerfreimnet AT benjaminhentze rotationalinvariantneuralnetworkforelectricalimpedancetomographyimagingwithoutreferencevoltagerfreimnet AT thomasmuders rotationalinvariantneuralnetworkforelectricalimpedancetomographyimagingwithoutreferencevoltagerfreimnet AT christianputensen rotationalinvariantneuralnetworkforelectricalimpedancetomographyimagingwithoutreferencevoltagerfreimnet AT steffenleonhardt rotationalinvariantneuralnetworkforelectricalimpedancetomographyimagingwithoutreferencevoltagerfreimnet AT chuongngo rotationalinvariantneuralnetworkforelectricalimpedancetomographyimagingwithoutreferencevoltagerfreimnet |