Abstract. We characterize the ability of electrical impedance tomography (EIT) to distinguish changes in internal conductivity distributions, and analyze it as a function of stimulation and measurement patterns. A distinguishability measure, z, is proposed which is related to the signal to noise ratio of a medium and to the probability of detection of conductivity changes in a region of interest. z is a function of the number of electrodes, the EIT stimulation and measurement protocol, the stimulation amplitude, the measurement noise, and the size and location of the contrasts. Using this measure we analyze various choices of stimulation and measurement patterns under the constraint of medical electrical safety limits (maximum current into the body). Analysis is performed for a planar placement of 16 electrodes for simulated 3D tank and chest shapes, and measurements in a saline tank. Results show that the traditional (and still most common) adjacent stimulation and measurement patterns have by far the poorest performance (by 6.9×). Good results are obtained for trigonometric patterns and for pair drive and measurement patterns separated by over 90 • . Since the possible improvement over adjacent patterns is so large, we present this result as a call to action: adjacent patterns are harmful, and should be abandoned. We recommend using pair drive and measurement patterns separated by one electrode less than 180 • . We describe an approach to modify an adjacent pattern EIT system by adjusting electrode placement.

Abstract. EIT can image the distribution of ventilated lung tissue, and is thus a promising technology to help monitor patient breathing to help selection of mechanical ventilation parameters. Two key difficulties in EIT instrumentation make such monitoring difficult: 1) EIT data quality depends on good electrode contact and is sensitive to changes in contact quality, and 2) EIT electrodes are difficult and time consuming to place on patients. This paper presents the design and initial tests of an active electrode based system to address these difficulties. Our active electrode EIT system incorporates an active electrode belt, a central voltage driven current source, central ADCs and DACs, a central FPGA based demodulator and controller. The electrode belt is designed incorporating 32 active electrodes, each of which contains the electronic amplifiers, switches and associated logic. Tests show stable device performance with a convenient ease of use and good imaging ability in volunteer tests.

Abstract. An electrical impedance tomography (EIT) system images internal conductivity from surface electrical stimulation and measurement. Such systems necessarily comprise multiple design choices from cables and hardware design to calibration and image reconstruction. In order to compare between EIT systems and to study the consequences of changes in system performance, this paper describes a systematic approach to evaluate the performance of the EIT systems. The system to be tested is connected to a saline phantom in which calibrated contrasting test objects are systematically positioned using a position controller. A set of evaluation parameters are proposed which characterize: i) data and image noise, ii) data accuracy, iii) detectability of single contrasts and distinguishability of multiple contrasts, and iv) accuracy of reconstructed image (amplitude, resolution, position and ringing). Using this approach, we evaluate three different EIT systems and illustrate the use of these tools to evaluate and compare performance. In order to facilitate use of this approach, all details of the phantom, test objects and position controller design are made publicly available including the source code of the evaluation and reporting software.

In this paper we propose a novel formulation for the distinguishability of conductivity targets in electrical impedance tomography (EIT). It is formulated in terms of a classic hypothesis test to make it directly applicable to experimental configurations. We test to distinguish conductivity distributions σ 2 from σ 1 , from which EIT measurements are obtained with added white Gaussian noise with covariance Σ n . In order to distinguish the distributions, we must reject the null hypothesis H 0 :x = 0, which has a probability based on the z-score: z =x σx . This result shows that distinguishability is a product of the impedance change amplitude, the measurement strategy and the inverse of the noise amplitude. This approach is used to explore different current stimulation strategies.

An automated test system and procedure is proposed, designed to enable systematic testing of electrical impedance tomography (EIT) devices. The system is designed to calculate reliable, repeatable and accurate performance figures of merit of an EIT system using a saline phantom and an industrial robot arm. Applications of the test system are to compare EIT devices against requirements, or to help optimize a device for its operating parameters. A test methodology and sample test results are presented to illustrate its use. The system is used to compare image quality and contrast detection for a range of stimulation and measurement patterns, and results show the best images when the pair of current injection electrodes is spaced between 45 and 170 degrees on a tank. Finally, we propose a classification of the object detection errors, which can facilitate comparison of EIT instrument specifications.