Modeling, system optimization and artifact reduction in magnetic induction tomography for medical applications

Magnetic induction tomography (MIT) is an emerging imaging modality and aims to reconstruct the electrical tissue properties in the region of interest. The noncontact and noninvasive measurement characteristics together with the fast scanning ability make the proposed modality attractive. It promises to facilitate diagnosis of several physiological disorders such as oedema or internal hemorrhage and has the potential to be used for the continuous monitoring of pathological fluid changes. This study focuses on improving the image quality and the investigation of several fundamental issues that must be tackled in order to have the modality as a diagnostic tool alongside clinical standard imaging modalities. To this end, several alternative coil configurations have suggested and the resulting imaging quality was compared with that of currently existing systems. Furthermore an optimization approach for the optimization of the coil design was developed. For this, the image performance measures are reviewed and a measure which favors the coil designs with the most independent information content is used to reach optimal designs. Besides the coil optimization, another important issue affecting image quality is the correction of patient positioning errors and movement artifacts in the numeric models. These errors were analyzed in simulation studies ands several solution strategies were proposed to compensate the artifacts. Lastly it was investigated to what extent the anisotropic electrical characteristics of the tissues must be considered in the forward and inverse models so as to keep reconstruction errors small. An anisotropic inverse solver was developed and the results were compared with those of a conventional isotropic solver. In particular conductivity tensor images were reconstructed to explore the effect of surface-near muscle tissue.