Search

We propose a measurement method using Walsh functions in 32-electrode electrical impedance tomography. By injecting currents in spatial Walsh function patterns, we use the simplicity of pulse type injection and yield the optimal SNR or distinguishability of sinusoidal injection. Once we measure boundary voltage data for the injection of all 31 patterns of Walsh functions, we can compute all 31 optimal current pattems. We can also synthesize the voltage data due to the injection of the optimal current patterns without actually injecting the optimal current patterns.

Static image reconstruction in electrical impedance tomopphy (EIT) using iterative algorithms requires a large computation time. We developed an improved NewtonRaphson (N-R) method which is more efficient, robust, and easily parallelizable. In order to check the feasibility of parallel computation in EIT, we implemented the algorithm on a parallel computer system. The results show a reduction in the computation time within a certain hiit as we increase the number of processors. We conclude that the improved N-R method is highly parallelizable and the use of parallel computers such as Transputers may solve the computational problems in static impedance imaging.

A 32-electrode, 12-bit electrical impedance tomograph (EIT) which is capable of injecting optimal patterns of currents into an object to achieve maximum distinguishability is presented. The software for the system is described, and static images reconstructed from a circular 2-D phantom are shown. The EIT software consists of four parts: the interactive graphical finite-element mesh generator; the sparse linear system equation solver including node renumbering schemes; the calculation of the optimal patterns of injection current and the implementations of the EIT image reconstruction algorithms; and image display and postprocessing. >