Correction of the non-uniform spatial sensitivity of electrical impedance tomography images.

One of the clinical applications of electrical impedance tomography (EIT) is the measurement of volume changes of body fluids. However, the spatial sensitivity of images produced by the Sheffield Mark I system is non-uniform and if quantitative comparisons of resistivity images resulting from volume changes are to be mad, this non-uniformity must be corrected for. All experimental data were collected from a phantom consisting of semi-permeable membrane tubes fixed vertically at different distances from the centre of a cylindrical tank which was filled with a saline solution of resistivity 5 omega m. Individual images were obtained when each tube was filled with a saline solution of resistivity 1.5 omega m. Analysis of these images, using the resistivity integral described by Thomas et al, yielded the sensitivity of EIT at different positions within the tank. Combinations of tubes were then filled with saline (1.5 omega m) and an image produced for each combination. Cross sections of the images were taken and used to confirm that the principle of superposition is valid for EIT. A simple pixel scaling correction, derived from the sensitivity data, was then applied to the images and the results re-analysed to demonstrate a reduction in volume measurement error. The application of this correction to in vivo images is discussed.