Long-term electrical impedance measurements are affected by specific errors. Electrode failure, changes in its impedance due to aging, and postural changes are among the most important. We analyse errors due to electrode replacement and body postural changes. Electrode replacement errors can cause impedance changes up to 5% of basal value. This is one of the most important factors in data reproducibility. Body postural changes also contribute to impedance variations. We have proposed the use of a reference position to carry out impedance measurements as the one that shows the smallest impedance sensitivity to postural changes. In general, we observed that this is achieved with arms and legs slightly separated from the body. We propose the use of a ratio of impedance at two different frequencies to discern the origin of impedance changes, whether from physiological phenomena or postural errors.

Static images of the human body using electrical impedance tomography techniques can be obtained by measuring at two or more different frequencies. The frequencies used depend on the application, and their selection depends on the frequency behaviour of the impedance for the target tissue. An analysis using available data and theoretical models for tissue impedance yields the expected impedance and boundary voltage changes, therefore setting the measurement instrument specifications. The instrument errors produced by different sources are analysed, and, from this analysis it is possible to determine the feasibility of building the instrument, the limit values for some parameters (or components) and indications on the most suitable design of critical parts. This analysis also shows what kinds of error can be expected in the reconstructed images. It is concluded that it is possible to build an instrument with limited errors, allowing static images to be obtained. An instrument has been built that meets some of the design requirements and fails in others because of technological problems. In vivo images obtained with this instrument will be presented in Part 2 of this work.

This paper deals with the design, implementation and performance of TIE-4sys, an electrical impedance tomograph. This instrument is a parallel broad-band real-time system. It measures impedance using an array of 16 electrodes and reconstructs the images using a weighted back-projection technique. The objective of this development is to enable multifrequency EIT clinical studies to be undertaken. The system is capable of acquiring 25 frames/s and makes multifrequency cardiac-gated images. The frequency range is from 10 kHz to 250 kHz and the signal to noise ratio for the real component is better than 60 dB.

A widely accepted method for static imaging in electrical impedance tomography (EIT) is to measure at two frequencies. The choice of measurement frequencies is application-dependent because some different tissues cannot be distinguished when using two fixed frequencies. We have developed a system that generates signals from 8-10(3) kHz and applies two of these signals simultaneously to the body through a broadband current mirror. Great care has been taken in the design of the current injection multiplexer in order to keep the current source output capacitance as low as possible. Furthermore design of the layout of the patient interface board, in order to reduce feedthrough capacitances, also needs great care. Other parameters for driving and detection sections have been designed according to our results from FEM and circuit simulations including skin and electrode effects. Simulations using FEM with available tissue impedance data and preliminary measurements in a discrete phantom show that static imaging is possible for both the real and imaginary parts of the impedance.