Yasin Mamatjan

An experimental clinical evaluation of EIT imaging with ℓ1 data and image norms

Electrical impedance tomography (EIT) produces an image of internal conductivity distributions in a body from current injection and electrical measurements at surface electrodes. Typically, image reconstruction is formulated using regularized schemes in which ℓ2-norms are used for both data misfit and image prior terms. Such a formulation is computationally convenient, but favours smooth conductivity solutions and is sensitive to outliers. Recent studies highlighted the potential of ℓ1-norm and provided the mathematical basis to improve image quality and robustness of the images to data outliers. In this paper, we (i) extended a primal-dual interior point method (PDIPM) algorithm to 2.5D EIT image reconstruction to solve ℓ1 and mixed ℓ1/ℓ2 formulations efficiently, (ii) evaluated the formulation on clinical and experimental data, and (iii) developed a practical strategy to select hyperparameters using the L-curve which requires minimum user-dependence. The PDIPM algorithm was evaluated using clinical and experimental scenarios on human lung and dog breathing with known electrode errors, which requires a rigorous regularization and causes the failure of reconstruction with an ℓ2-norm solution. The results showed that an ℓ1 solution is not only more robust to unavoidable measurement errors in a clinical setting, but it also provides high contrast resolution on organ boundaries.

Enhancing Impedance Imaging Through Multimodal Tomography

Several noninvasive modalities including electrical impedance tomography (EIT), magnetic induction tomography (MIT), and induced-current EIT (ICEIT) have been developed for imaging the electrical conductivity distribution within a human body. Although these modalities differ in how the excitation and detection circuitry (electrodes or coils) are implemented, they share a number of common principles not only within the image reconstruction approaches but also with respect to the basic principle of generating a current density distribution inside a body and recording the resultant electric fields. In this paper, we are interested in comparing differences between these modalities and in theoretically understanding the compromises involved, despite the increased hardware cost and complexity that such a multimodal system brings along. To systematically assess the merits of combining data, we performed 3-D simulations for each modality and for the multimodal system by combining all available data. The normalized sensitivity matrices were computed for each modality based on the finite element method, and singular value decomposition was performed on the resultant matrices. We used both global and regional quality measures to evaluate and compare different modalities. This study has shown that the condition number of the sensitivity matrix obtained from the multimodal tomography with 16-electrode and 16-coil is much lower than the condition number produced in the conventional 16-channel EIT and MIT systems, and thus, produced promising results in terms of image stability. An improvement of about 20% in image resolution can be achieved considering feasible signal-to-noise ratio levels.

Imaging of hemorrhagic stroke in magnetic induction tomography: An in vitro study

Magnetic Induction Tomography (MIT) has the potential of providing an inexpensive medical device for regular screening and monitoring of patients. MIT could be used to detect hemorrhagic stroke, if high measurement accuracy and spatial resolution can be reached to allow significant contrast between normal brain tissues and hemorrhaged areas. However, this can be challenging because spatial resolution in MIT is limited due to the small number of independent measurements, the inverse problems are severely ill‐posed, and erroneous data cause large artefacts in reconstructed images and lower the detectability threshold of bleeding. The noise components degrading signal and image quality may be caused by thermal drift and noise from acquisition systems, environment or by body movements. The objective of the article is to empirically investigate hemorrhagic stroke in MIT based on in vitro study and to improve stroke detectability and visibility to help monitoring stroke patients. The following approaches were evaluated: (i) level setting, (ii) improved spatial filtering, (iii) averaging of multiple measurements, (iv) the combinations of these three approaches, and (v) wavelet denoising. They were evaluated with an in vitro phantom resembling a cerebral stroke in a pig brain. The results showed that these approaches enhanced stroke visibility, lowered stroke detectability threshold from 15 ml to 5 ml, and improved the localization of phantom hemorrhages such that their combination produced the best results. These methods may make it easy to the estimation of actual stroke volume, and clinical interpretation, and it can be used to long‐term monitoring of stroke progression.

Evaluation and Real-Time Monitoring of Data Quality in Electrical Impedance Tomography

Electrical impedance tomography (EIT) is a noninvasive method to image conductivity distributions within a body. One promising application of EIT is to monitor ventilation in patients as a real-time bedside tool. Thus, it is essential that an EIT system reliably provide meaningful information, or alert clinicians when this is impossible. Because the reconstructed images are very sensitive to system instabilities (primarily from electrode connection variability and movement), EIT systems should continuously monitor and, if possible, correct for such errors. Motivated by this requirement, we describe a novel approach to quantitatively measure EIT data quality. Our goals are to define the requirements of a data quality metric, develop a metric q which meets these requirements, and an efficient way to calculate it. The developed metric q was validated using data from saline tank experiments and a retrospective clinical study. Additionally, we show that q may be used to compare the performance of EIT systems using phantom measurements. Results suggest that the calculated metric reflects well the quality of reconstructed EIT images for both phantom and clinical data. The proposed measure can thus be used for real-time assessment of EIT data quality and, hence, to indicate the reliability of any derived physiological information.

Electrical localization of weakly electric fish using neural networks

Weakly Electric Fish (WEF) emit an Electric Organ Discharge (EOD), which travels through the surrounding water and enables WEF to locate nearby objects or to communicate between individuals. Previous tracking of WEF has been conducted using infrared (IR) cameras and subsequent image processing. The limitation of visual tracking is its relatively low frame-rate and lack of reliability when visually obstructed. Thus, there is a need for reliable monitoring of WEF location and behaviour. The objective of this study is to provide an alternative and non-invasive means of tracking WEF in real-time using neural networks (NN). This study was carried out in three stages. First stage was to recreate voltage distributions by simulating the WEF using EIDORS and finite element method (FEM) modelling. Second stage was to validate the model using phantom data acquired from an Electrical Impedance Tomography (EIT) based system, including a phantom fish and tank. In the third stage, the measurement data was acquired using a restrained WEF within a tank. We trained the NN based on the voltage distributions for different locations of the WEF. With networks trained on the acquired data, we tracked new locations of the WEF and observed the movement patterns. The results showed a strong correlation between expected and calculated values of WEF position in one dimension, yielding a high spatial resolution within 1 cm and 10 times higher temporal resolution than IR cameras. Thus, the developed approach could be used as a practical method to non-invasively monitor the WEF in real-time.

Experimental/clinical evaluation of EIT image reconstruction with ℓ1 data and image norms

Electrical impedance tomography (EIT) image reconstruction is ill-posed, and the spatial resolution of reconstructed images is low due to the diffuse propagation of current and limited number of independent measurements. Generally, image reconstruction is formulated using a regularized scheme in which l2 norms are preferred for both the data misfit and image prior terms due to computational convenience which result in smooth solutions. However, recent work on a Primal Dual-Interior Point Method (PDIPM) framework showed its effectiveness in dealing with the minimization problem. l1 norms on data and regularization terms in EIT image reconstruction address both problems of reconstruction with sharp edges and dealing with measurement errors. We aim for a clinical and experimental evaluation of the PDIPM method by selecting scenarios (human lung and dog breathing) with known electrode errors, which require a rigorous regularization and cause the failure of reconstructions with l2 norm. Results demonstrate the applicability of PDIPM algorithms, especially l1 data and regularization norms for clinical applications of EIT showing that l1 solution is not only more robust to measurement errors in clinical setting, but also provides high contrast resolution on organ boundaries.

A Novel Method for Monitoring Data Quality in Electrical Impedance Tomography

Electrical impedance tomography (EIT) has the promise to help improve care for patients undergoing ventilation therapy by providing real-time bed-side information on the distribution of ventilation in their lungs. To realise this potential, it is important for an EIT system to provide a reliable and meaningful signal at all times, or alert clinicians when this is not possible. Because the reconstructed images in EIT are sensitive to system instabilities (including electrode connection problems) and artifacts caused by e.g. movement or sweat, there is a need for EIT systems to continuously monitor, recognize and, if possible, correct for such errors. Motivated by this requirement, our paper describes a novel approach to quantitatively measure EIT data quality suitable for online and offline applications. We used a publicly available data set of ventilation data from two pediatric patients with lung disease to evaluate the data quality on clinical data. Results suggest that the developed data quality could be a useful tool for real-time assessment of the quality of EIT data and, hence, to indicate the reliability of any derived physiological information.