Robustness of Single Ended Measurement to Electrode Errors in Electrical Impedance Tomography: An Experimental Study

Abstract

Detachment of electrodes is the most common problem in EIT while monitoring due to patient's movement or sweat etc. which results in significant image artifacts. Though several compensation strategies had been proposed to manage these errors, however it also degrades the quality of actual image especially if the target is closer to the electrodes. The degradation effect not only depends on the compensation strategy but also on the measurement methodology used. It has already been reported earlier that for the adjacent method (differential measurement) the actual images are not significantly affected if the target is at a distance greater than 15% of the diameter from the affected electrode. Recently single ended measurements with internal electrode have gained more popularity because of its greater sensitivity. This paper reports on the robustness of single ended measurement towards electrode errors. In vitro studies are carried out for single ended measurement(common ground method) and differential measurement(adjacent method) to find how the actual images are affected after compensating for electrode errors as the target moves closer to the affected electrode and as the number of failed electrodes increases. Adjacent electrodes are removed to have the maximum effect on the target as the target is placed closer to the affected electrodes. Different algorithms had been suggested earlier to detect these faulty electrodes in case of differential measurements. In this paper we have also discussed the applicability of current voltage reciprocity principle to automatically detect faulty electrodes in case of single ended measurement. Maximum number of adjacent and non-adjacent faulty electrodes that can be detected for both methods using the algorithm is reported. It is observed that for single ended measurement up to four adjacent faulty electrodes are detected however, in case of adjacent method only up to two faulty electrodes are detected. Bayesian imaging model, where electrode errors are fed as a priori large measurement noise on all measurements using affected electrode is used to account for the erroneous data. In case of single ended measurement there isn't any significant change in resolution and position error of the reconstructed image for 1 faulty electrode if the target is at a distance greater than or equal to 10% of the diameter from the electrodes. Also for single ended measurement the images are not significantly affected even for 3 adjacent faulty electrodes after compensation. Overall, results indicate that single ended measurement is more robust and gives better images than differential measurement after compensating for electrode errors.