Current patterns and electrode types for single-source electrical impedance tomography of the thorax.

Abstract

Electrical impedance tomography (EIT) estimates the spatial distribution of the electrical tissue properties in a cross section of the body. In the present study, we investigated how the quality of static thoracic images obtained from EIT systems with a single current source and sink is affected by the current pattern employed in the presence of measurement noise. The reconstructed images best reproduced our computational phantom when current source and sink were placed at neighboring electrodes. In this case, the mean squared reconstruction error was an order of magnitude smaller than for all other patterns of current injection studied. At a signal-to-noise ratio of 50 dB, 60% of the reconstructions converged successfully with source and sink at neighboring electrodes, while only 10% or less converged for all other configurations. We relate these results to the fact that neighboring currents strengthen the diagonal structure in the Hessian matrix of the iterative reconstruction process that we employed. We also tested the effects on the reconstruction error of the number and type of electrodes. We found that "compound electrodes" that permit voltage measurement at the site of current injection did not yield any practical improvement of the image quality. In contrast, doubling the number of boundary electrodes reduced the reconstruction error by almost two orders of magnitude.