In multi-frequency electrical impedance tomography (EIT) systems it is much easier to design voltage sources than current sources. It is also at times desirable to operate a voltage driving system as if it were a current driving system. This can be done by adjusting the individual voltage sources until a desired current pattern is obtained. Questions remain regarding the form of the adjustment algorithm and the circumstances under which it will converge. Through simulation and experimentation we have developed a simple algorithm which functions satisfactorily in most practical situations. We have also investigated its theoretical limits for convergence. Simulations showed that convergence is reached in all cases with little (0.1%) or no measurement noise. With moderate noise (0.5%) our algorithm failed to converge in 17 out of 6000 runs, while more significant measurement noise (1%) resulted in convergence in only 7 out of 6000 runs. Experiments with a 16-channel EIT system converged in all cases on volunteer arms, but failed some of the time in saline, with the number of successful runs decreasing with frequency (73.3% at 10 kHz, 71.1% at 125 kHz and 15.5% at 750 kHz), suggesting a possible link to measurement error.
[1]
Keith D. Paulsen,et al.
A multichannel continuously selectable multifrequency electrical impedance spectroscopy measurement system
,
2000,
IEEE Transactions on Biomedical Engineering.
[2]
Keith D. Paulsen,et al.
Simulation of error propagation in finite element image reconstruction for electrical impedance tomography
,
2001
.
[3]
M. Cheney,et al.
Distinguishability in impedance imaging
,
1992,
IEEE Transactions on Biomedical Engineering.
[4]
P Bertemes-Filho,et al.
A comparison of modified Howland circuits as current generators with current mirror type circuits.
,
2000,
Physiological measurement.