Research On Image Reconstruction Algorithm Based On Medical Electrical Impedance Tomography

With advantages over other CT techniques, e.g. non-intrusion, portability, low cost and fast response, Electrical Impedance Tomography (EIT) has been investigated extensively in the last decade, and found many important applications in both industrial and medical areas.In EIT, an array of electrodes is attached around the object and alternating currents are injected via these electrodes and the resulting voltages are measured. Using the measurements on the boundary, an approximation for the spatial distribution of impedance within the object can be reconstructed. Mathematically, the physical model for EIT meets the Laplace elliptic partial differential equation. Since there is no analytical solution for an arbitrary impedance distribution in the volume, Finite Element Method (FEM) is chosen to solve the forward problem in EIT. The EIT reconstruction problem is a nonlinear ill-posed inverse problem, the solution of which requires the use of regularization methods.In this thesis, the forward problem in EIT is solved by using COMSOL Multiphysics@ with Matlab@ which can be modeled flexibly and solved accurately. The forward problem of EIT is solved combined with human thorax models of two/three dimension, which are imported by COMSOL from CT scanning. Basic theories of inverse problem are discussed. Besides, in order to overcome the shortage of data, several reconstruction methods based on prior information are introduced:1. Based on traditional method of Tikhonov regularization, incorporate the system smoothness information by using regularization matrix, which improves the resolution of reconstruction image. Furthermore, import maximum entropy punish function into generalized Tikhonov method, which makes advantage of its non-negative characteristics to enhance the image reconstruction of non-negative conductivity distribution in practical problems.2. Utilizing CT scanning, build up human thorax model, combined with prior information, such as the structure of tissues and conductivity distribution of different tissues. A reconstruction method based on thorax model is proposed, which has improved the quality of thorax reconstruction image effectively.3. Considering the changes of tissue structure and conductivity distribution due to the cycle of respiration, the basis constrained method is introduced, based on a learning process from which a few basis images are obtained and the solution is forced to be a linear combination of these few images. In the learning process the available prior information that is based on the object properties is effectively utilized. Due to the parameter reduction the reconstruction procedure becomes fast. However, it was found that if the actual conductivity distribution is in accordance with the prior this method gives good results, but if it is not, it will deal with misleading results.4. Based on the Tikhonov regularization technique, a modification of the basis constrained method is discussed. In this method the solution is drawn towards the assumed solution, that is, towards the linear combination of the basis images with the aid of properly constructed regularization matrix. This method gives good resistivity estimates when the actual conductivity distribution is in accordance with the prior but even if it is not, the conductivity distribution can be also well reconstructed.5. Establish three dimensional thorax model and make some elementary research.Finally, the author presents some ideas to improve the medical EIT image reconstruction.