Study On Numerical Algorithm And Fundmental Technology Of3D Electrical Impedance Tomography

Electrical Impedance Tomography（EIT） is a new medical imaging modality thatimages by computing electrical properties inside the human body. In the clinicalmedical field it has a good application prospect, such as gastrointestinal disease, lungdisease, cancer diagnosis and brain disease. Compared to the current medical imagingtechnology, EIT has its own advantages, for example, functional imaging, simpleequipment, low-cost and non-invasion. So far the study for EIT has focused on the2-Dmodel. The electrode system of current3D EIT is multilayer closed electrode which isused in the lung, stomach disease diagnosis. While in the breast cancer andintra-abdominal ooze blood disease diagnosis, the conductivity distribution under theskin of the shallow area need to be study. This paper have study the3D EIT include themeasurement system, the stimulatied model, the reconstruction algorithm and physicalexperiment. This system can be used in breast cancer diagnosis and abdominal bleedingdetection. The main contents of the dissertation includethe following parts:1) The current breast cancer detection equipments, including the domestic andforeign existing EIT systems, were studied and analysed. The structure of plateelectrodes and the back electrode was proposed for clinical application, which makesthe current distribution more uniform, as well as increases the detecting depth.2) As for the forward problem of3-D EIT, the finite element equations of completeelectrode model were deduced based on the3-D electrode modeling and themathematical model in the electromagnetic problem. Finite element programs werecompiled in Matlab, and commercial electromagnetic field simulation software Comsolwas applied to compare the accuracy of the programs. The results show that the errorbetween Matlab and Comsol is less than4%, which indicate that Matlab programs canbe used as the basis of inverse calculation. The simulations of finite element programswas used to contrast several existing stimulation patterns, then two methods, thedistinguishability and surface projection image methods, were put forward to evaluatethe stimulation pattern. The results show that the back electrode pattern has obviousadvantages in differentiation degree and the projective quality, which provides providereference for the system design.3) Baesd on the characteristics of the inverse problems of EIT, the Tikhonovregularization algorithm and Newton’s One-step Error Reconstructor (NOSER) were used to reconstructed the3D field firstly. The optimal method which is called L-curvemethod was used to select the better Tikhonov regularization parameter. And then theregularization matrix which is similar to the Laplace operator was also given. Then, a3-D cylinder model is set for simulation. The results show that the single regularizationalgorithm has its own merits and demerits, and therefore a hybrid algorithm waspresented based on Tikhonov regularization and the NOSER.Finally, a series ofsimulations were carried out to compare these three algorithms.The encouraging resultsindicate that this proposed algorithm can reduce the number of condition, make thereconstruction process stable and improve the spstial resolution. Owing to the greatamount of calculation in3D inverse problem, based on the2D image reconstruction foropen EIT,3D interpolation imaging algorithm was put forward, and linear interpolationalgorithm was used to get the images between two2D layers.4) The hardware system was developed. Based on the deep research on the keyproblem of EIT measurement and constant flow source, the principle and realization of3D multi-frequency measurement system were introduced. The experiment prototypewas built and the basic performances had been tested.5) To test the effectiveness of the performance of the prototype and the combinedreconstruction algorithm, the physical model experiment were carried out, includingtank dynamic imaging experiment, agar dynamic model experiment and the frequencydifference imaging experiments. A simulation experiment of abdominal bleeding wasdesigned, by changing the clot volume to simulate the different blood loss. Theexperimental results show that this3D imaging system can detect the1-cm diametertarget in4-cm-deep and distinguish different volume of blood clots. The spatialresolution of the image and the contrast is better.The final part is the summary description. This paper summarized the majorresearch achievements, as well as the existing problems and deficiencies. Meanwhile,the further researches on this topic were previewed.