Image Reconstruction Using Supervised Learning in Wearable Electrical Impedance Tomography of the Thorax

Abstract

Electrical impedance tomography (EIT) is a non-invasive technique for visualizing the internal structure of a human body. Capacitively coupled electrical impedance tomography (CCEIT) is a new contactless EIT technique that can potentially be used as a wearable device. Recent studies have shown that a machine learning-based approach is very promising for EIT image reconstruction. Most of the studies concern models containing up to 22 electrodes and focus on using different artificial neural network models, from simple shallow networks to complex convolutional networks. However, the use of convolutional networks in image reconstruction with a higher number of electrodes requires further investigation. In this work, two different architectures of artificial networks were used for CCEIT image reconstruction: a fully connected deep neural network and a conditional generative adversarial network (cGAN). The training dataset was generated by the numerical simulation of a thorax phantom with healthy and illness-affected lungs. Three kinds of illnesses, pneumothorax, pleural effusion, and hydropneumothorax, were modeled using the electrical properties of the tissues. The thorax phantom included the heart, aorta, spine, and lungs. The sensor with 32 area electrodes was used in the numerical model. The ECTsim custom-designed toolbox for Matlab was used to solve the forward problem and measurement simulation. Two artificial neural networks were trained with supervision for image reconstruction. Reconstruction quality was compared between those networks and one-step algebraic reconstruction methods such as linear back projection and pseudoinverse with Tikhonov regularization. This evaluation was based on pixel-to-pixel metrics such as root-mean-square error, structural similarity index, 2D correlation coefficient, and peak signal-to-noise ratio. Additionally, the diagnostic value measured by the ROC AUC metric was used to assess the image quality. The results showed that obtaining information about regional lung function (regions affected by pneumothorax or pleural effusion) is possible using image reconstruction based on supervised learning and deep neural networks in EIT. The results obtained using cGAN are strongly better than those obtained using a fully connected network, especially in the case of noisy measurement data. However, diagnostic value estimation showed that even algebraic methods allow us to obtain satisfactory results.

Generative-Adversarial-Network-Based Image Reconstruction for the Capacitively Coupled Electrical Impedance Tomography of Stroke

“…Comparing the quality measures obtained on the brain image dataset with the results obtained on the thorax dataset [ 15 ], it is essential to notice that brain reconstruction appears to be a more complicated task for the neural network. This was expected since the brain is surrounded by a skull with a much lower conductivity and electrical permittivity.…”

“…Our previous work [ 15 ] showed that modified cGAN-based neural network architecture inspired by the Pix2Pix approach [ 50 ] is very promising for electrical tomography image reconstruction. Previously, this network was trained on the imaginary component of normalized measurement data obtained by simulating measurements for an empty sensor and the sensor filled with high-conductivity material.…”

“…The cGAN approach is widely used in EIT image reconstruction. We have already successfully confirmed the possibility of using cGAN in thoracal CCEIT [ 15 ]. However, according to our knowledge, there have been no attempts to explore its feasibility for human brain CCEIT image reconstruction.…”

“…In EIT, four-electrode measurement limits the influence of electrode impedance on the measurement, but compensation is insufficient if an electrode impedance mismatch occurs [ 10 , 11 , 12 ]. In CCEIT, a two-electrode measurement is performed in the form of electrical capacitance tomography (ECT) [ 13 , 14 , 15 ]. The contactless electrodes are insulated from the skin by a dielectric and do not require gluing to the skin with gel.…”

Multifrequency Electrical Impedance Tomography Images for Swine Pneumonia Detection Using Conductivity Differences

“…Both monofrequency and Multifrequency Electrical Impedance Tomography techniques have been developed since the 1980s and serve various applications in medicine. These include monitoring and imaging in cases of hospitalization due to SARS-CoV-2 [1][2][3], acute respiratory distress syndrome (SARS) [4], lung pathologies using artificial intelligence [5], lung cancer [6], lung embolism [7], breast cancer imaging [8,9], prostate tumors [10], tumor detection [11], and emphysema [12]. MfEIT has also been used in conjunction with other imaging modalities, such as magnetic resonance imaging, magnetic induction tomography, and computed tomography, to enhance the speed and quality of image acquisition.…”