SOURCES OF ERROR IN BIOIMPEDANCE SPECTROSCOPY

Two different makes of bioimpedance spectrometer (UniQuest-SEAC SFB3 a nd Xitron 4000B) were used for a series of measurements on volunteers and patients in intensive care. Although each machine was accurate ove r the frequency range 5 to 500 kHz when bench tested on model resistor -capacitor circuits, significant differences in their recorded impedan ce parameters appeared when used in vivo, especially on intensive care patients. A series of laboratory tests was performed on each machine simulating the situation in vivo to identify possible reasons for thes e differences. Whilst stray capacitance in the environment was identif ied as the major contributor to variability in high-frequency performa nce, interaction between electrode impedance and lead positioning was also a factor. The observed phase shift with frequency or time delay ( T-d) used in the Xitron modelling software appears to be the result of a time constant caused by stray capacitance and so is unlikely to hav e any biological meaning. Significant differences in the in vivo numer ical values produced by bioimpedance spectrometers may be attributed t o instrument design, data processing and, in particular, the clinical environment.