Magnetic Resonance Electrical Impedance Tomography (MREIT) reconstructs images of electrical conductivity distribution based on magnetic flux density (B) measurements. Magnetic flux density is generated by an externally applied current on the object and measured by a Magnetic Resonance Imaging (MRI) scanner. With the measured data and peripheral voltage measurements, the conductivity distribution of the object can be reconstructed. There are two types of reconstruction algorithms. First type uses current density distributions to reconstruct conductivity distribution. Object must be rotated in MRI scanner to measure three components of magnetic flux density. These types of algorithms are called J-based reconstruction algorithms. The second type of reconstruction algorithms uses only one component of magnetic flux density which is parallel to the main magnetic field of MRI scanner. This eliminates the need of subject rotation. These types of algorithms are called B-based reconstruction algorithms. In this study four of the B-based reconstruction algorithms, proposed by several research groups, are examined. The algorithms are tested by different computer models for noise-free and noisy data. For noise-free data, the algorithms work successfully. System SNR 30, 20 and 13 are used for noisy data. For noisy data the performance of algorithm is not as satisfactory as noise-free data. Twice differentiation of z component of B (Bz) is used for two of the algorithms. These algorithms are very sensitive to noise. One of the algorithms uses only one differentiation of Bz so it is immune to noise. The other algorithm uses sensitivity matrix to reconstruct conductivity distribution.

Magnetic Resonance Electrical Impedance Tomography (MREIT) reconstructs conductivity distribution with internal current density (MRCDI) and boundary voltage measurements. There are many algorithms proposed for the solution of MREIT inverse problem which can be divided into two groups: Current density (J) and magnetic flux density (B) based reconstruction algorithms. In this thesis, J-based MREIT reconstruction algorithms are implemented and optimized with modifications. These algorithms are simulated with five conductivity models which have different geometries and conductivity values. Results of simulation are discussed and reconstruction algorithms are compared according to their performances. Equipotential-Projection algorithm has lower error percentages than other algorithms for noise-free case whereas Hybrid algorithm has the best performance for noisy cases. Although J-substitution and Hybrid algorithms have relatively long reconstruction times, they produced the best images perceptually. v Integration along Cartesian Grid Lines and Integration along Equipotential Lines algorithms diverge as noise level increases. Equipotential-Projection algorithm has erroneous lines starting from corners of FOV especially for noisy cases whereas Solution as a Linear Equation System has a typical grid artifact. When performance with data of experiment 1 is considered, only Solution as a Linear Equation System algorithm partially reconstructed all elements which show that it is robust to noise. Equipotential-Projection algorithm reconstructed resistive element partially and other algorithms failed in reconstruction of conductivity distribution. Experimental results obtained with a higher conductivity contrast show that Solution as a Linear Equation System, J-Substitution and Hybrid algorithms reconstructed both phantom elements and Hybrid algorithm is superior to other algorithms in percentage error comparison.

Electrical conductivity of biological tissues is a distinctive property which differs among tissues. It also varies according to the physiological and pathological state of tissues. Furthermore, in order to solve the bioelectric field problems accurately, electrical conductivity information is essential. Magnetic Resonance Electrical Impedance Tomography (MREIT) technique is proposed to image this information with high spatial resolution. However, almost all MREIT algorithms proposed to date assumes isotropic conductivity in order to simplify the underlying mathematics. But it is known that most of the tissues in human body have anisotropic conductivity values. The aim of this study is to reconstruct anisotropic conductivity images with MREIT. In the study, five novel anisotropic conductivity reconstruction algorithms are developed and implemented. Proposed algorithms are grouped into two: current density based reconstruction algorithms (Type-I) and magnetic flux density based algorithms (Type-II). Performances of the algorithms are evaluated in several aspects and compared with each other. The technique is experimentally realized using 0.15T METU &ndash / EE MRI System and anisotropic conductivity images of test phantoms are reconstructed using all proposed algorithms.

Breast cancer is the most crucial cancer type among all other cancer types. There are many imaging techniques used to screen breast carcinoma. These are mammography, ultrasound, computed tomography, magnetic resonance imaging, infrared imaging, positron emission tomography and electrical impedance tomography. However, there is no gold standard in breast carcinoma diagnosis. The object of this study is to create a hybrid system that uses thermal and electrical imaging methods together for breast cancer diagnosis. Body tissues have different electrical conductivity values depending on their state of health and types. Consequently, one can get information about the anatomy of the human body and tissue&rsquo / s health by imaging tissue conductivity distribution. Due to metabolic heat generation values and thermal characteristics that differ from tissue to tissue, thermal imaging has started to play an important role in medical diagnosis. To increase the temperature contrast in thermal images, the characteristics of the two imaging modalities can be combined. This is achieved by implementing thermal imaging applying electrical currents from the body surface within safety limits (i.e., thermal imaging in active mode). Electrical conductivity of tissues changes with frequency, so it is possible to obtain more than one thermal image for the same body. Combining these images, more detailed information about the tumor tissue can be acquired. This may increase the accuracy in diagnosis while tumor can be detected at deeper locations. Feasibility of the proposed technique is investigated with analytical and numerical simulations and experimental studies. 2-D and 3-D numerical models of the female breast are developed and feasibility work is implemented in the frequency range of 10 kHz and 800 MHz. Temporal and spatial temperature distributions are obtained at desired depths. Thermal body-phantoms are developed to simulate the healthy breast and tumor tissues in experimental studies. Thermograms of these phantoms are obtained using two different infrared cameras (microbolometer uncooled and cooled Quantum Well Infrared Photodetectors). Single and dual tumor tissues are determined using the ratio of uniform (healthy) and inhomogeneous (tumor) images. Single tumor (1 cm away from boundary) causes 55 &deg / mC temperature increase and dual tumor (2 cm away from boundary) leads to 50 &deg / mC temperature contrast. With multi-frequency current application (in the range of 10 kHz-800 MHz), the temperature contrast generated by 3.4 mm3 tumor at 9 mm depth can be detected with the state-of-the-art thermal imagers.

This dissertation focuses on the application of tomography in the sugar milling process, specifically within the vacuum pan. The research aims to improve the efficiency and throughput of a sugar mill by producing real-time images of the boiling dynamic in the pan and hence can be used as a diagnostic tool. The real-time tomography system is a combination of ruggedized data collecting hardware, a switching circuit and software algorithms. The system described in this dissertation uses 16 electrodes and estimates images based on the distinct differences in conductivities to be found in the vacuum pan, i.e. a conductive syrup-like fluid (massecuite) and bubbles. There is a direct correlation between the bubbles produced during the boiling process and heat transfer in the pan. From this correlation one can determine how well the pan is operating. The system has been developed in order to monitor specific parts of a pan for optimal boiling. A binary reconstructed image identifies either massecuite or water vapour. Each image is reconstructed using a modified neighbourhood data collection method and a back projection algorithm. The data collection and image reconstruction take place simultaneously, making it possible to generate images in real-time. Each image frame is reconstructed at approximately 1.1 frames per second. Most of the system was developed in LabVIEW, with some added external drive electronics, and functions seamlessly. The tomography system is LAN enabled hence measurements are initiated through a remote PC on the same network and the reconstructed images are streamed to the user. The laboratory results demonstrate that it is possible to generate tomographic images from bubbles vs massecuite, tap water and deionized water in real-time. / Thesis (M.Sc.Eng.)-University of KwaZulu-Natal, Durban, 2012.

Magnetic resonance imaging (MRI) is a high resolution medical imaging technique based on distinguishing tissues according to their nuclear magnetic properties. Magnetic resonance electrical impedance tomography (MREIT) is a conductivity imaging technique which reconstructs images of electrical properties, based on their effect on induced magnetic flux density due to externally applied current flow. Both of these techniques are of interest for novel research and development. Simulators help researchers observe the accuracy and the results of the study. In this study a user friendly complete MRI/MREIT simulator is designed. This simulator is the combination of improved version of MRI simulator (implemented by V. E. Arpinar, H. Yigitler), a forward solver, to observe the current injection effect, the improved version of user interface that is designed on LabVIEW graphical programming environment (designed by M. Ozsut), and equi-potential projection (EPP) reconstruction algorithm (proposed by M. S. Ozdemir, M. Eyuboglu, O. Ozbek). All of these individual parts are improved and gathered in LabVIEW environment in order to work in synchrony. In addition to that, regional image reconstruction technique (proposed by H. Altunel, M. Eyuboglu) is also included in the simulator. The simulator is run for various inputs and system specifications. It is observed that the simulation results are consistent with the expected results for MRI, MREIT and conventional/regional MREIT reconstruction. Four different models are designed and results are obtained using these models. The accuracy of the results usually differs with the input parameters and model geometry. Validating numerically the accuracy of the forward solution part using Biot-Savart and Ampere&#039 / s laws, the consistency of the forward problem solution part is obtained at a percentage of 95%. In the MREIT part, magnetic flux density distribution taken from forward solver part is added to the main magnetic flux density used in the MRI part. Consistency of the magnetic flux density distribution given to the simulator as input and the output taken from the MREIT part of the simulator is found as 99%. In addition to conventional EPP algorithm, regional MREIT reconstruction algorithm is applied for various noise levels. It is observed that, as the noise level increases, regional MREIT reconstruction algorithm gives relatively much better results compared to conventional MREIT reconstruction algorithm. Errors obtained by applying conventional reconstruction and regional reconstruction are compared for each inhomogeneity individually. Therefore, accuracies of the different current patterns depending on the inhomogeneities are observed as well.

abstract: Among electrical properties of living tissues, the differentiation of tissues or organs provided by electrical conductivity is superior. The pathological condition of living tissues is inferred from the spatial distribution of conductivity. Magnetic Resonance Electrical Impedance Tomography (MREIT) is a relatively new non-invasive conductivity imaging technique. The majority of conductivity reconstruction algorithms are suitable for isotropic conductivity distributions. However, tissues such as cardiac muscle and white matter in the brain are highly anisotropic. Until recently, the conductivity distributions of anisotropic samples were solved using isotropic conductivity reconstruction algorithms. First and second spatial derivatives of conductivity (∇σ and ∇2σ ) are integrated to obtain the conductivity distribution. Existing algorithms estimate a scalar conductivity instead of a tensor in anisotropic samples. Accurate determination of the spatial distribution of a conductivity tensor in an anisotropic sample necessitates the development of anisotropic conductivity tensor image reconstruction techniques. Therefore, experimental studies investigating the effect of ∇2σ on degree of anisotropy is necessary. The purpose of the thesis is to compare the influence of ∇2σ on the degree of anisotropy under two different orthogonal current injection pairs. The anisotropic property of tissues such as white matter is investigated by constructing stable TX-151 gel layer phantoms with varying degrees of anisotropy. MREIT and Diffusion Magnetic Resonance Imaging (DWI) experiments were conducted to probe the conductivity and diffusion properties of phantoms. MREIT involved current injection synchronized to a spin-echo pulse sequence. Similarities and differences in the divergence of the vector field of ∇σ (∇2σ) among anisotropic samples subjected to two different current injection pairs were studied. DWI of anisotropic phantoms involved the application of diffusion-weighted magnetic field gradients with a spin-echo pulse sequence. Eigenvalues and eigenvectors of diffusion tensors were compared to characterize diffusion properties of anisotropic phantoms. The orientation of current injection electrode pair and degree of anisotropy influence the spatial distribution of ∇2σ. Anisotropy in conductivity is preserved in ∇2σ subjected to non-symmetric electric fields. Non-symmetry in electric field is observed in current injections parallel and perpendicular to the orientation of gel layers. The principal eigenvalue and eigenvector in the phantom with maximum anisotropy display diffusion anisotropy. / Dissertation/Thesis / Masters Thesis Bioengineering 2015

Electrical Impedance Tomography (EIT) is a technique for imaging the spatial distribution of conductivity inside a body using the boundary voltages, in response to applied current patterns, to reconstruct an image. Even though EIT has been proved useful in several medical applications such as mechanical respiration and ventilation monitoring of the lungs, its reported success in localising cerebral conductivity changes due to brain stimulation is very scant. In the case of the human head, the amplitude of the brain response to stimulation is usually very small and gets contaminated with physiological noise initiated from inside the cranium or the scalp. Three types of evoked responses were experimentally investigated: auditory startle response (ASR), CO2 reactivity response, and transient hyperaemic response (THR). ASR is expected to be a result of the brain’s functioning processes. However, the responses to CO2 and THR are expected to be due to cerebral blood volume or flow, due to physiological intervention in blood supply. According to the results, even when the amplitude of EIT measurements shows profound variation as in the case of CO2 reactivation, those could not be physiologically linked to the targeted responses and have been shown to be initiated from the scalp. The consistency of the measurements in the case of CO2 reactivation response was poor (37.50-50%). Meanwhile in the case of THR, although the magnitude of conductivity changes was overall 50% smaller than the previous cases, the subject movement was not necessary. This could be a reason that the consistency of THR case was very good (87%), and this can emphasize the necessity to maintain the changes in the scalp at minimum levels. In the case of ASR the response magnitude was very small (six times smaller than the CO2 reactivity case), and the evoked response can be detected with only 50% consistency. To measure very small EIT signals (such as those expected due to brain function) effectively, one must improve the sensitivity of the measurements to conductivity changes by increasing the excitation current. The functional EIT for Evoked Response (fEITER) system used in our investigations was modified from its initial configuration to increase its excitation current from 1 mApk-pk to 2 mApk-pk or 1 mArms. The bit-truncation in the process of Phase-Sensitive Detection (PSD) has also been improved, to modify the original 16-bit data readout to be 24-bit data readout. These improvements have doubled the instrument’s sensitivity, and have substantially reduced the truncation error to about 183 times. The quality of the physiological waveform was also significantly improved. Therefore, one could study more effectively very fast brain response using the modified system. For example, the latency of responses can be more precisely extracted, or the monitoring of the conductivity change in a period of only a few tens of milliseconds is then possible. The reconstruction of brain images corresponding to these physiologically evoked responses has been the ultimate goal of this thesis. To ensure obtaining the correct images, some crucial issues regarding EIT reconstruction were firstly investigated. One of these issues concerns the modelling error of the numerical head models. The reconstruction requires an accurate model capturing the geometry of the subject’s head with electrodes attached and accurate in-vivo tissue conductivities. However, since it is usually impractical to have a personalised model for each subject, many different head models (including a subject model) were constructed and investigated, to evaluate the possibility of using a generic model for all subjects. The electrode geometry was also carefully included into the models to minimise error. Another issue concerns the appropriate reconstruction algorithm. A novel nonlinear reconstruction method, based on the difference imaging approach and Generalized Minimal Residual method (GMRes) algorithm, with optimal parameters and prior information, was proposed to deal with significant modelling errors. With this algorithm, the experimental results showed that it is possible to use a generic model for reconstructing an impedance change, but the magnitude of the change should be rather small. The last issue tackled was regarding the a priori choice of model parameters, and in particular the tissue conductivities. The tissue conductivities of the scalp and the skull were also estimated by a proposed methodology based on the Gauss-Newton method. The estimation showed that, compared to previous reported values, the conductivity of the scalp was higher, at 0.58 S/m, and that of the skull lower, at 0.008 S/m. Eventually, by exploiting the hardware and firmware advances in the measuring instrument in conjunction with the proposed modelling and reconstruction algorithm, processing our experimental EIT data captured on human heads and a head-like tank confirm that the localisation and imaging of conductivity changes occurring within the head is indeed possible. From the low quality measurements in the case of the CO2 reactivity response, the reconstructed images of this response do not reflect the true conductivity change. The consistency of the images to localise the sources of the changes was very poor (0-50%), i.e. the conductivity changing locations in the images were likely to be random. Our analysis suggests that the changes inside the cranium are likely to be due to the large change in the scalp. In the case of THR, the reconstructed images were able to localise the response in a similar manner to what had been found on the measurements, and the consistency was quite high (76%). Meanwhile, in the case of ASR, surprisingly the consistency of the images was 82%, much higher than the consistency of the measurements, which was only 50%. This was because the changing amplitude of the measurements was too small to be noticed by visualisation, and it was practically cumbersome to investigate all measurements. This statistic confirms that image reconstruction can reveal information that is not directly apparent by observing the measurements. In summary, EIT can be used in brain (function) imaging applications to some extent. The targeted response, which typically originates from inside the cranium is always infused with neurophysiological noise or physical noise at the scalp, and the amplitude of noise determines the possibility to localise the changes. It is also necessary for the desired response to have sufficiently large amplitude. These results show that EIT has been successful in THR and ASR, but for CO2 reactivity response, EIT lacks the necessary sensitivity.

Atualmente existem diversos métodos e equipamentos disponíveis no mercado para análise da biomecânica do movimento humano. No entanto, mesmo uma equipe multidisciplinar, dispondo de um laboratório completo de análise do movimento, pode falhar na identificação de quais grupos musculares estão sendo recrutados durante um exercício. Sobretudo quando a musculatura de interesse é profunda. O objetivo desta tese é propor formas de detectar a contração muscular através da Tomografia por Impedância Elétrica (TIE). Um modelo de elementos finitos de condução elétrica é utilizado para resolver o problema inverso através do algoritmo de Newton-Raphson de forma a obter as imagens de TIE. Um novo modelo de eletrodo e o método de erro de discretização da malha são introduzidos como forma de melhorar as imagens de TIE. Além disso, a variabilidade da impeditividade de tecidos musculo-esqueléticos é medida experimentalmente, in vivo tanto em repouso quanto em exercício. Os resultados mostram que o sangue tem um papel importante nas mudanças de impeditividade e que as variações medidas durante as contrações musculares parecem estar relacionadas à taxa de contração do movimento. As imagens de TIE, obtidas in vivo de um voluntário, apresentam um aumento de resistividade durante a contração muscular. / Presently, there are several methods and equipment available in the market for the biomechanical analysis of human movement. However, even a well trained multidisciplinary team, equipped with a complete motion analysis laboratory, may fail to identify which muscle groups are being recruited during an exercise. Specially when deep muscles are being considered. The main objective of this work is to propose forms to detect muscle contraction from Electrical Impedance Tomography (EIT) images. A finite element electrical conduction model is used to solve an inverse problem with Newton-Raphson algorithm in order to produce EIT images. A new electrode model is proposed and the mesh discretization error method is implemented to improve EIT images. Additionally the variability of impeditivity of musculo-skeletal tissues is measured experimentally in vivo both at rest and during exercise. The results show that blood has an important role in muscle impeditivity changes and that resistivity variations during muscle contractions seem to be related to movement contraction rate. The EIT images, obtained in vivo from a volunteer, show an increase of resistivity during muscle contraction.

Le toucher est considéré comme l’un des sens primordiaux à modéliser chez un robot afin de lui permettre de générer des comportements plus souples et plus agiles comme attraper un objet, toucher (ou être touché par) une personne. Même si les capteurs tactiles actuels sont encore très limités en comparaison à la peau humaine, combinés à la vision et à la proprioception, le développement de nouveaux capteurs proches de la peau humaine pourrait démultiplier les capacités d’interactions d’un robot afin d’interagir directement avec une personne en toute sécurité et de partager avec lui son environnement physique et social. A la différence de la peau humaine, les principaux capteurs tactiles utilisés en robotique actuellement ne sont capables de détecter des variations de pression et de poids que sur de petites surfaces uniquement. De plus, ceux-ci sont souvent très rigides et n’ont pas les propriétés élastiques de déformation de la peau humaine. Les travaux de cette thèse se basent sur le développement d’une interface tactile proche d’une "peau artificielle" en terme de surface de recouvrement (qui peuvent atteindre plusieurs dizaines de centimètres carrés) et de localisation des points de contact de quelques dizaines de millinewtons. Deux aspects principaux sont développés : (i) aspect d’ingénierie comprenant le développement d’un prototype de peau artificielle conçue pour un robot humanoïde afin de lui conférer une perception tactile, et (ii) aspect cognitifs qui s’appuient sur l’intégration de multiples rétroactions sensorielles (tactile, visuelle, proprioceptive) dans le but d’avoir un robot qui puisse interagir physiquement avec des personnes.Le prototype tactile développé est basé sur la reconstruction du champ électrique à la surface d’un matériau conducteur, suivant le principe de la Tomographie par Impédance Électrique (TIE). Notre innovation principale a été d’implémenter des techniques d’apprentissage par réseau de neurones artificiels afin de reconstruire l’information sans utiliser les techniques analytiques d’inversion de matrice coûteuse en temps de calcul. De plus, nous montrons que l’utilisation de réseaux de neurones artificiels permet d’avoir un système beaucoup plus biomimétique, indispensable pour comprendre la perception du toucher chez l’être humain.Nous avons ensuite abordé le problème de l’intégration des informations tactiles et motrices. Après avoir recouvert un bras manipulateur avec la peau artificielle, nous avons fait apprendre un réseau de neurones son schéma corporel et adapter sa compliance par retour tactile. Le fonctionnement du moteur est basé sur le contrôle par admittance du bras robotique. Des expériences montrent que les réseaux de neurones peuvent contrôler l’interaction adaptative entre le bras du robot avec une personne grâce à l’estimation du couple appris selon la position où la force tactile avait été appliquée lors de la phase d’apprentissage.Enfin, nous nous sommes intéressées à la problématique de la représentation du corps au niveau neuronal, comment les êtres humains perçoivent leur propre corps à travers tous les sens (visuel, tactile et proprioceptif). Nous avons proposé un modèle biologique au niveau du cortex pariétal qui s’appuie sur l’intégration de multiples rétroactions sensorielles du corps du robot (son bras) et sur la synchronisation des rétroactions visuelles et proprioceptives. Nos résultats montrent l’apprentissage d’une image corporelle et l’espace péri-personnel avec l’émergence de neurones qui codent une information spatiale visuo-tactile relative au déplacement du bras et centrée soit sur le bras robotique soit centrée sur l’objet. / The touch perception is considered as one of the crucial senses to be recreated in a robot so that it could generate a more flexible and agile behavior. For instance, grasping an object, as well as touch or be touched by a person. Although modern touch sensors are still very limited compared to the human skin, combined with vision and proprioception, the development of new sensors similar to human skin could multiply the robot’s capacity to interact directly and safely with a person, as well as to share his or her physical and social environment.Unlike human skin, the main touch sensors used in modern robotics are only capable of detecting the pressure and weight variations on small batches of surface. Moreover, they are often quite stiff and do not have the elastic deformation capacity intrinsic to the human skin. The purpose of this thesis is to develop a touch interface close to "artificial skin" in terms of the covered area (which can reach several square decimeters) and localization of the contact points (several dozen millinewtons). Two main aspects have been developed: (i) the engineering aspect including the development of an artificial skin prototype for a humanoid robot designed to impart a tactile perception, and (ii) the cognitive aspect that is based on the integration of multiple sensory feedbacks (tactile, visual, proprioceptive) in order to conceive a robot that can physically interact with people.The developed tactile prototype is based on the reconstruction of the electric field on the surface of a conductive material, following the principle of Electrical Impedance Tomography (EIT). Our main innovation was to implement the neural network learning techniques to reconstruct the information without using the inverse matrix analytical techniques which imply time consuming computation. Moreover, we show that the application of artificial neural networks allows to obtain a much more biomimetic system, essential to understand the perception of the human touch.Then, we addressed the issue of integrating tactile and motor information. After having covered a manipulator arm with artificial skin, we have learn a neural network its body schema and enables it to adjust its compliance with tactile feedback. The functioning of the motor is based on the admittance control of the robot arm. Experiments show that neural networks can control the adaptive interaction between the robot arm and a human being by estimating the torque perceived according to the position where the touch force had been applied during the learning phase.Finally, we turned our attention to the issue of the body representation at the neuronal level, namely, how human beings perceive their own body through all their senses (visual, tactile, and proprioceptive). We have proposed a biological model in the parietal cortex, which is based on the integration of multiple sensory feedbacks from the robot’s body (its arm) and on the synchronization of visual and proprioceptive feedback. Our results show the capacity to perceive the body image with the emergence of neurons that encode a spatial visual-tactile information of the arm movement and is centered on either the robotic arm or on the object.