IMPROVED METHODS TO DETERMINE OPTIMAL CURRENTS IN ELECTRICAL-IMPEDANCE TOMOGRAPHY

Our electrical impedance tomography (EIT) system uses the optimal current method to inject currents and the regularized Newton-Raphson algorithm to reconstruct an image of resistivity distribution. The optimal current patterns, however, are a function of the unknown resistivity distribution, thus they cannot be obtained analytically. In this paper we developed iterative methods to derive the optimal current patterns through iterative physical measurements. We also developed direct methods to first determine the resistance matrix of a resistivity distribution through a set of experimental measurements, then used the singular value decomposition (SVD) to obtain the optimal current patterns. In both the iterative and direct methods, we injected a complete set of current bases and stored the measured voltage responses. This permitted iterative reconstruction techniques to operate on the stored data without requiring lengthy data taking from the object. This reduced the effects or motion artifacts. We concluded that the direct methods have superior performance as compared to the iterative methods in both optimal current and voltage generation. We studied three sets of current bases: Fourier, diagonal, and neighboring. The Fourier-based, method produced most accurate results but required multiple current generators. The diagonal-based method produced slightly less accurate but comparable results using the simple hardware of a single current generator.