A statistical approach to the inclusion of electrode contact impedance uncertainty in electrical tomography reconstruction

Electrical tomography is a visualisation tool used for industrial process monitoring. The complete electrode model relates an unknown conductivity field with the measurements but also involves unknown electrode contact impedances. Here, a real data analysis shows that the contact impedances vary spatially and with time. Then, the main reconstruction process is repeated using contact impedance values drawn at random from a fitted contact impedance distribution, and a model average calculated as the final reconstruction. This additional source of variation can then be appropriately accounted for thereby preventing overly optimistic assumption in subsequent decision making.