Research On Reconstruction Algorithm Of Magnetic Resonance Electrical Resistance Tomography

Magnetic Resonance Electrical Resistance Tomography(MRERT), which combines the traditional Electrical Resonance Tomography(ERT) and the Magnetic Resonance Imaging(MRI) technique, is a new non-intrusive imaging modality. The obvious advantage of MRERT is to use MRI to detect the inner information of a multiphase flow within a cross-section under the influence of injecting a current from the boundary of object being measured. The measurements from MRI can be used to reconstruct the physical distribution of the imaged cross-section with an improved image resolution comparing to those of images obtained by ERT measurement due to that the amount of information is increased significantly in MRERT measurement. Likewise the problem of ill-condition in the matrix of reconstruction as a result of lacking independent measurements information in ERT is also mitigated in MRERT image reconstruction. At present, MRERT has attracted extensive attentions worldwide in applications of biomedicine, geophysics, industrial detection and so on.In this thesis, ERT as an imaging measurement technique for multiphase flow measurement is reviewed and its current development status and existing problems are summarized. MRERT as a measurement technique are then introduced to highlight its significant technical advantages compared with the imaging quality that are obtainable by an ERT system, followed with a brief review of major researches and applications of MRERT in home and abroad. It then goes to technical details on MRERT system development, both forward and inverse solvers, the FEM modeling of multiphase flow and electric-magnetic field distribution, the potential and current density distributions, the quantitative calculation of Lamor frequency shift in each pixel as a result of current disturbance.Also in this thesis a sensitivity matrix based approach is adopted to give a qualitative reconstruction of the resistivity distribution in a 2D space. Unlike in ERT measurement where multiple injecting of current is often necessary to achieve a desired image resolution, in MRERT it generally requires only one injecting of current to yield sufficient information for image reconstruction. A series of simulations of using sensitivity algorithm on flow patterns, measurement noise effects, multiphase spatial distributions and gap distance in between the oil droplets are performed to study the performance of MRERT. Tikhonov regularization is used to improve the image quality of the sensitivity algotithm. Simulation shows that MRERT can significantly improve the image quality and resolution compared with what can be achieved by ERT. Although the results achieved are preliminary, the work lays an essential foundation on future development of an MRERT system for industrial process measurement.