Effects of head model inaccuracies on regional scalp and skull conductivity estimation using real EIT measurements

Abstract

we estimate the scalp and skull conductivities on two healthy adults, based on bounded (or parametric) Electrical Impedance Tomography (bEIT) measurements, and using 62 current injection pairs of a high dense 128 sensor array. We compare the estimates obtained with three different electrode models: pointwise, volumetric, and the Complete Electrode Model (CEM). We also analyze the influence of the skull details and the cerebrospinal fluid (CSF). The estimated scalp (skull) conductivities for these two subjects were ~0.4 and ~0.3 S/m (~0.0045 and ~0.005 S/m), similar for all three electrode models (within 8%). Volumetric and CEM models resulted in a better fit to real data. A model of nested and closed surfaces (no skull holes) resulted in a significant overestimation (~23%) of the skull conductivity. Moreover, neglecting the CSF resulted in an extra 28% overestimation of the skull conductivity. This clearly shows the need of precise head modeling for bEIT.