CompanyName: Techn. Univ. of Graz, IGTE
Abstract: EIT spectroscopy by direct multifrequency-regularization of the Gauss-Newton method
B. Brandstätter, K. Hollaus, Ch. Magele, M. Mayer, R. Merwa, H. Scharfetter and H. Hutten
A major drawback of EIT is the poor
quality of the conductivity images, i.e., the low spatial resolution as well
as large errors in the reconstructed conductivity values due to the low sensitivity
far away from the objects surface. The main reason is the necessity for regularization
of the ill-conditioned inverse problem which results in excessive low-pass filtering.
A novel regularization method based on the direct inclusion of multi frequency information is proposed. Up to now EIT is usually performed by reconstructing the conductivity distribution for several frequencies and a-posteriori fitting of a tissue model (e. g. Cole model). The new approach, in contrast, exploits the additional a-priori information contained in the shape of typical tissue spectra directly for the reconstruction. The unknowns are the parameters of an appropriate tissue model (e. g. Cole parameters) instead of the conductivities. Therefore an additional constraint for the dependence of the measured data at different frequencies is introduced. As regularization term a smoothness criterion is imposed on the conductivity distribution as calculated from the tissue model.
The iteration scheme is a modified Gauss-Newton method, wherein the Jacobian of the mapping between parameters and the surface voltages is calculated with the Geselowitz theorem. Main advantage of this approach is the optimization of the accuracy of the (physiological) tissue parameters of interest instead of the conductivities. The performance of the algorithm will be compared with classical reconstruction approaches for a two dimensional finite element model.