Machine Learning Approach to EIT Image Reconstruction of the Human Forearm Section for Different Hand Signs

Aug 5, 2026

Electrical impedance tomography (EIT) reconstructs an object’s internal conductivity from boundary voltage measurements. This study evaluates image reconstruction algorithms for the human forearm by inferring muscle contractions associated with different hand signs. Because EIT image reconstruction is an ill-posed inverse problem, Gauss-Newton often requires many iterations to tune regularization and compute Jacobians. To reduce computation, machine learning is used for direct reconstruction, comparing a Radial Basis Neural Network (RBNN) and a 1D convolutional neural network (1D-CNN). Trained on measured EIT data from eight subjects, both methods achieve low deviation (0.0017 for RBNN, 0.0109 for CNN).

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages8

Reading Time20 min

AccessFree to download

Introduction

Motivation for EIT and ill-posed reconstruction

Hand signs for gesture recognition

Methodology

Hand sign data

Reconstruction algorithms and evaluation metrics

Results

Reconstruction performance for experimental data

Conclusion

Future work

MACHINE LEARNING APPROACH TO EIT IMAGE RECONSTRUCTION OF THE HUMAN FOREARM SECTION FOR DIFFERENT HAND SIGNS HayatAlasasfeh, Mariem Hafsa, Oumaima Bader, and Olfa Kanoun Professorship of Measurement and Sensor Technology, Technische Universitat Chemnitz, Chemnitz, Germany Université de Sousse, Ecole Nationale d’Ingénieurs de Sousse, LATIS-Laboratory of Advanced Technology and Intelligent Systems, 4023, Sousse, Tunisie; ABSTRACT Electrical impedance tomography (EIT) is an imaging technique used to reconstruct the conductivity of a target object from boundary voltages. In this study, we investigate suitable image reconstruction algorithms for EIT to enable the reconstruction of the conductivity distribution in the forearm section inferring muscle contractions at different hand signs. As EIT image reconstruction is an ill-posed inverse problem, the Gauss-Newton algorithm needs many iterations for the determination of suitable values of the regularization parameter and corresponding calculations of the Jacobian matrix. To reduce computational effort, we propose to use machine learning algorithms to directly reconstruct the EIT image. We explore the Radial Basis Neural Network (RBNN) and a one-dimensional Convolutional Neural Network (1DCNN), which has been trained based on the measured EIT data for eight subjects, ten hand signs with ten trials. Both methods reach a low deviation at 0.0017 for RBNN and 0.0109 for CNN. Index Terms – Image reconstruction, EIT, CNN, RBNN, Hand signs 1. INTRODUCTION EIT is a non-invasive imaging technique that uses electrical currents to reconstruct the internal conductivity distribution of an object [1]. EIT is a radiation-free and cost-effective technology with applications in diverse fields such as medical imaging, geological exploration, industrial process monitoring, and environmental studies [2] . In EIT, low-amplitude electrical currents are injected between two electrodes on the surface of an object, and the resulting voltages are measured between other electrodes. These measurements are used to reconstruct a 2D or 3D image. Due to its portability and safety, EIT is increasingly being used in clinical filed such as monitoring lung ventilation, cerebral hemodynamics, breast cancer detection, and gesture recognition [3] . Hand signs (HSs) are a nonverbal communication approach in which the human hand expresses emotions or information. HSs are increasingly important in technology, particularly in humancomputer interaction. With the proliferation of touchscreens, motion sensors, and augmented reality devices, it has become an intuitive and natural way to interact with technology [4] . In healthcare, it's being used to develop more intuitive and non-invasive interfaces for medical devices, enabling doctors and surgeons to control equipment and medical data. As such, the potential for revolutionizing the way to interact with technology and making it more accessible and intuitive for everyone lies within the development of HSs technology[5] . Different movements in the wrist move the internal muscle tissue and bones, which alters the conductivity distribution in the forearm. EIT can be used to reconstruct the conductivity distribution of the forearm because the EIT electrodes are unaffected by the external environment [6] . © 2023 by the authors.–Licensee Technische UniversitätIlmenau, Deutschland. Intelligent Algorithms can also be an alternative to traditional reconstruction algorithms, which are always considered a regression problem [7]. Therefore, it can be classified into evolutionary methods and Neural Networks (NN) . Evolutionary methods primarily include the Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) [8][9] by lowering the root mean relative square error between simulated and measured data. The evolutionary algorithms are difficult to attain high precision, and the computation process is time-consuming. The NN method avoids the linearization of EIT im

FAQs

What problem does this study address in EIT image reconstruction?

EIT reconstruction is an ill-posed inverse problem, so traditional Gauss-Newton approaches require many iterations to tune the regularization parameter and compute Jacobian-related quantities.

Which machine learning models are compared for forearm conductivity reconstruction?

The study compares a Radial Basis Neural Network (RBNN) and a one-dimensional convolutional neural network (1D-CNN), both trained on measured EIT data.

How is the hand-sign data used in the reconstruction task?

Measured EIT data are collected from eight subjects across ten hand signs with ten trials each, and the trained models output conductivity vectors corresponding to different gestures.

Aug 5, 2026

Upload by Olivia Brown