Study On Forword Problem And Reconstruction Algorithm In Magnetic Induction Tomography

The imaging technique of magnetic induction tomography (MIT) is a new method to measure the conductivity distribution of biological tissue according to electromagnetic detection principle. MIT technique is no ray radiation, small, non-contact, functional, real-time clinical monitoring. Image reconstruction is the final imaging method of MIT, the application of MIT technology will be directly affected by the reconstruction algorithm's precision and imaging speed. The present magnetic induction tomography is not mature enough to meet the clinical standards; the main technique problems include high-accuracy measuring system, high-resolution and fast–converged reconstruction algorithms. According to reconstruction algorithms, the paper mainly researches the calculation and simulation of forward problem and also researches the ill-posed amendment and rapid calculation of reconstruction algorithm.According to theory of MIT, The paper obtains the finite element equation for solution the forward problem by applying electromagnetic theory and uses Galerkin finite element method to solve the equation. The simulation experiment adopts imaging model with 20cm's diameter and 40 coils uniform distribution in the circle. The paper program the forward problem, The simulation results indict: The imaginary part of magnetic flux density and real part of eddy current density can reflect the location information of perturbation; The phase shift of detecting coil increases as it closed to object or is far away from exciting coil; To the same position of the object, the phase shift in detecting coil is linear to the conductivity, which provides prove to imaging reconstruction.According to reconstruction algorithm's ill-pose and Newton-Raphson algorithm characteristic, The paper applies eigenvalue threshold to modify NOSER which is abridged for NOSER-ET. Compared with the reconstructed results using TSVD and Tikhonov regularization algorithms, The NOSER-ET can obtain a better image quality with higher resolution. By adding gauss noise to phase deviation data, the NOSER-ET algorithm can also distinguish the objects, but the error of traditional algorithms are big. Through the above process, the algorithm in this paper improves the image accuracy and anti-noise characteristic; the algorithm also can locate accurately position of the complex imaging models; because the algorithm has no iterative procedure, so it enhances imaging speed and provides foundation for MIT technology clinical application.