On Optimal Current Patterns for Electrical Impedance Tomography

Abstract

We develop a statistical criterion for optimal patterns in planar circular electrical impedance tomography. These patterns minimize the total variance of the estimation for the resistance or conductance matrix. It is shown that trigonometric patterns (Isaacson, 1986), originally derived from the concept of distinguishability, are a special case of our optimal statistical patterns. New optimal random patterns are introduced. Recovering the electrical properties of the measured body is greatly simplified when optimal patterns are used. The Neumann-to-Dirichlet map and the optimal patterns are derived for a homogeneous medium with an arbitrary distribution of the electrodes on the periphery. As a special case, optimal patterns are developed for a practical EIT system with a finite number of electrodes. For a general nonhomogeneous medium, with no a priori restriction, the optimal patterns for the resistance and conductance matrix are the same. However, for a homogeneous medium, the best current pattern is the worst voltage pattern and vice versa. We study the effect of the number and the width of the electrodes on the estimate of resistivity and conductivity in a homogeneous medium. We confirm experimentally that the optimal patterns produce minimum conductivity variance in a homogeneous medium. Our statistical model is able to discriminate between a homogenous agar phantom and one with a 2 mm air hole with error probability (p-value) 1/1000.

The reconstruction problem

“…In this Bayesian framework the choice of optimal current patterns depends on the prior and a good choise will result in a 'tighter' posterior. Demidenoko et al [47] consider optimal current patterns in the framework conventional optimal design of experiments, and define an optimal set of current patterns as one that minimizes the total variance of Z.…”

Towards the Development of an EIT-based Stretchable Sensor for Multi-Touch Industrial Human-Computer Interaction Systems

“…For this reason, recent studies [6][7][8] have focused on developing different drive patterns for current excitation and voltage reading in order to improve the image resolution and object distinguishability. Being a drive pattern a strategy that selects the electrodes pairs on which current injection and voltage reading are performed, it affects system performance as different current injections result in different current density distribution across the transducer, and different voltage readings influence how big of a portion of the material is been read.…”

Research on Optimal Excitation and Measurement Mode of Electrical Resistance Tomography

“…Excitation mode refers to how the exciting current is injected into the test field, and common modes [7] include adjacent method, cross method and relative method, which are all based on the dual-electrode excitation method. Jiang et al put forward a single electrode excitation method [8], in which electrode current is calculated by detecting the voltage of the standard resistance that connected to the electrode; Demidenko et al proposed adaptive optimal excitation method [9], where a plurality of electrodes are stimulated simultaneously to obtain the best current distribution. Measurement mode refers to the voltage measurement method of boundary electrode, and general measurement modes involve adjacent measurement and reference measurement [10].…”