%0 Journal Article %1 1742-6596-224-1-012145 %A Peters, Franciane C %A Barra, Luis Paulo S %D 2010 %J Journal of Physics: Conference Series %K inverse problems %N 1 %P 012145 %T Some numerical results on the influence of measurement strategies and load patterns in the EIT inverse problem %U http://stacks.iop.org/1742-6596/224/i=1/a=012145 %V 224 %X It is well known that the Electrical Impedance Tomography reconstructed images are very sensitive to noise in the electrical potential measures at the external boundary of the domain. Furthermore, there are a variety of strategies to inject current and measure these values. Thus, the aim of this work is to verify how much some of these strategies affect the quality of the reconstructed image and the sensitivity of each strategy to the existence of noise in the measured data. For a known position and shape of a single homogeneous inclusion inside a conductor domain, measured data is numerically simulated by the forward problem solution via Boundary Element Method (BEM). Sixteen electrodes placed at the outer boundary of the domain are used with different load patterns and measurement strategies to obtain the exact data for the numerical experiments. In order to simulate noisy data, Additive White Gaussian Noise proportional to the exact measured value is used on each potential measure. The inverse problem is treated as a data fitting problem solved via Levenberg-Marquardt method, that minimizes the difference between measures data and computed potential values. The coordinates of the control points of a spline function that defines the boundary of the inclusion are used as minimization parameters. In each step of this procedure, the computed potential values are obtained via BEM. The identification of inclusions of different sizes, shapes and positions is presented allowing some conclusions about the influence of the studied parameters.

@article{1742-6596-224-1-012145, abstract = {It is well known that the Electrical Impedance Tomography reconstructed images are very sensitive to noise in the electrical potential measures at the external boundary of the domain. Furthermore, there are a variety of strategies to inject current and measure these values. Thus, the aim of this work is to verify how much some of these strategies affect the quality of the reconstructed image and the sensitivity of each strategy to the existence of noise in the measured data. For a known position and shape of a single homogeneous inclusion inside a conductor domain, measured data is numerically simulated by the forward problem solution via Boundary Element Method (BEM). Sixteen electrodes placed at the outer boundary of the domain are used with different load patterns and measurement strategies to obtain the exact data for the numerical experiments. In order to simulate noisy data, Additive White Gaussian Noise proportional to the exact measured value is used on each potential measure. The inverse problem is treated as a data fitting problem solved via Levenberg-Marquardt method, that minimizes the difference between measures data and computed potential values. The coordinates of the control points of a spline function that defines the boundary of the inclusion are used as minimization parameters. In each step of this procedure, the computed potential values are obtained via BEM. The identification of inclusions of different sizes, shapes and positions is presented allowing some conclusions about the influence of the studied parameters.}, added-at = {2010-05-24T17:05:49.000+0200}, author = {Peters, Franciane C and Barra, Luis Paulo S}, biburl = {https://www.bibsonomy.org/bibtex/2cd709c94ef6f0f12dbc7f11fe2ab7431/fcpeters}, interhash = {eaec9775e18e9e3e1bc0f43a507caf36}, intrahash = {cd709c94ef6f0f12dbc7f11fe2ab7431}, journal = {Journal of Physics: Conference Series}, keywords = {inverse problems}, number = 1, pages = 012145, timestamp = {2010-05-24T17:05:49.000+0200}, title = {Some numerical results on the influence of measurement strategies and load patterns in the EIT inverse problem}, url = {http://stacks.iop.org/1742-6596/224/i=1/a=012145}, volume = 224, year = 2010 }