Contactless Electrical Impedance Tomography with Deep Learning for Lung Monitoring: Phantom Study

This study proposes the application of contactless electrical impedance tomography (CEIT) for lung imaging, which is a crucial tool for respiratory disease diagnosis and monitoring. By employing the contactless measurement approach, CEIT not only avoids the contact impedance of traditional EIT but also has the potential to reduce discomfort for users, such as avoiding pressure on the chest caused by traditional EIT electrode belts. This study designed a phantom experiment to investigate the feasibility of CEIT in monitoring lung respiration. The deformations of the lung simulators are monitored through the difference in magnitude and phase angle of the measured impedance data. An image reconstruction algorithm combining the Landweber iteration and neural network was proposed. The method comprehensively utilized impedance magnitude and phase angle information, significantly improving the quality of CEIT's image reconstruction. With the proposed image reconstruction method, our CEIT can effectively achieve ventilation monitoring. It is a significant advancement for CEIT in the field of lung monitoring.

I. Introduction

Lung monitoring is of great significance for the diagnosis and post-treatment of patients in the intensive care unit and those suffering from respiratory obstructions. Electrical Impedance Tomography (EIT) has significant advantages over commonly used medical imaging techniques like CT and MRI, such as being non-invasive, radiation-free, and cost-effective. In the field of medical imaging, particularly in lung monitoring, EIT holds substantial research importance. It is widely used for bedside monitoring of global and regional lung function [1] and for bedside monitoring of lung ventilation [2]. EIT reconstructs the conductivity distribution of the human body's sensing region by applying current to the body surface through electrodes and measuring the corresponding voltages [3]. The direct contact between EIT electrodes and the human body introduces contact impedance, and the skin-electrode contact impedance is highly sensitive to body movements [4], [5]. This presents a challenge for the medical application of EIT. To mitigate the impact of contact impedance, this study would propose the use of contactless electrical impedance tomography (CEIT) for lung monitoring which avoids the contact impedance by utilizing a contactless measurement method. Yandan Jiang was the first to investigate the application of CEIT in biological imaging/biomedicine through biological tissue phantom experiments [6]. The application of CEIT in the medical field also includes feasibility studies on brain stroke imaging [7] and breast cancer detection [8], both of which have been conducted through phantom experiments. [9] investigated the feasibility of CEIT with deep learning in imaging local lung function through numerical simulation. Research on the application of CEIT in the biological field is relatively limited but holds significant research significance. In the monitoring of lung respiration, CEIT has potential applications. This study investigates the feasibility of using CEIT for monitoring lung respiration through a phantom experiment. The image reconstruction algorithm combining the Landweber method and neural network is proposed to enhance the image reconstruction quality of CEIT. Section II introduces the principle of CEIT and the setup of the phantom experiment. Section III presents the framework of the proposed image reconstruction algorithm and Section IV shows the experimental results.