Study of the optimum level of electrode placement for the evaluation of absolute lung resistivity with the Mk3.5 EIT system.

Inter-subject variability has caused the majority of previous electrical impedance tomography (EIT) techniques to focus on the derivation of relative or difference measures of in vivo tissue resistivity. Implicit in these techniques is the requirement for a reference or previously defined data set. This study assesses the accuracy and optimum electrode placement strategy for a recently developed method which estimates an absolute value of organ resistivity without recourse to a reference data set. Since this measurement of tissue resistivity is absolute, in Ohm metres, it should be possible to use EIT measurements for the objective diagnosis of lung diseases such as pulmonary oedema and emphysema. However, the stability and reproducibility of the method have not yet been investigated fully. To investigate these problems, this study used a Sheffield Mk3.5 system which was configured to operate with eight measurement electrodes. As a result of this study, the absolute resistivity measurement was found to be insensitive to the electrode level between 4 and 5 cm above the xiphoid process. The level of the electrode plane was varied between 2 cm and 7 cm above the xiphoid process. Absolute lung resistivity in 18 normal subjects (age 22.6 +/- 4.9, height 169.1 +/- 5.7 cm, weight 60.6 +/- 4.5 kg, body mass index 21.2 +/- 1.6: mean +/- standard deviation) was measured during both normal and deep breathing for 1 min. Three sets of measurements were made over a period of several days on each of nine of the normal male subjects. No significant differences in absolute lung resistivity were found, either during normal tidal breathing between the electrode levels of 4 and 5 cm (9.3 +/- 2.4 Omega m, 9.6 +/- 1.9 Omega m at 4 and 5 cm, respectively: mean +/- standard deviation) or during deep breathing between the electrode levels of 4 and 5 cm (10.9 +/- 2.9 Omega m and 11.1 +/- 2.3 Omega m, respectively: mean +/- standard deviation). However, the differences in absolute lung resistivity between normal and deep tidal breathing at the same electrode level are significant. No significant difference was found in the coefficient of variation between the electrode levels of 4 and 5 cm (9.5 +/- 3.6%, 8.5 +/- 3.2% at 4 and 5 cm, respectively: mean +/- standard deviation in individual subjects). Therefore, the electrode levels of 4 and 5 cm above the xiphoid process showed reasonable reliability in the measurement of absolute lung resistivity both among individuals and over time.

Detection of emboli in vessels using electrical impedance measurements--phantom and electrodes.

A phantom was constructed to simulate the electrical properties of the neck. A range of possible electrode configurations was then examined in order to improve the sensitivity of the impedance measurement method for the in vivo detection of air emboli. The neck phantom consisted of simulated skin, fat and muscle layers made of agar and a conductive rubber tube mimicking the common carotid artery. The ring-shaped electrodes with a guard electrode showed the highest sensitivity to emboli at short distances.

Electrical bioimpedance readings increase with higher pressure applied to the measuring probe.

Electrical bioimpedance spectroscopy (EBIS) is a technique that uses a probe to calculate the transfer impedance from tissues. This transfer impedance can give information about the normal or pathological condition of the tissue. To take readings, pressure has to be applied to the probe in order to get a good contact between the electrodes and the tissue. We have been using EBIS to investigate the early diagnosis of dysplasia and cancer in the human cervix, oesophagus and bladder. We have found that, with increasing pressure (range used here was approximately 1 kPa to approximately 50 kPa), the resistivity readings increase in a consistent way up to 80%. In this paper, we show how this is a case in three different tissue types (oesophageal, gastric and vesical samples). These increases can be higher than those associated with the pathological changes that we are investigating (non-inflamed columnar tissue, for instance, shows values 50% higher than dysplastic columnar tissue). Finite-element modelling was also used to investigate the effect of volume reduction in the connective tissue or stroma. This simulation suggests no strong correlation between reduction of this structure and increase in resistivity. We hypothesize therefore that these changes may be mainly associated with the squeezing of water from the extracellular space. Finally, as pressure is difficult to control by hand, we raise the issue of the necessity of considering this variable when making EIS measurements.

Modelling the electrical properties of bladder tissue--quantifying impedance changes due to inflammation and oedema.

Electrical impedance spectroscopy has been developed as a potential method for the diagnosis of carcinoma in epithelial tissues. An understanding of the influence of structural changes in the tissue on the properties measured using this technique is essential for interpreting measured data and optimization of probe design. In contrast to other tissue types, carcinoma in situ of the bladder gives rise to an increase in electrical impedance over the kHz-MHz frequency range in comparison to normal tissue. Finite element models of the urothelium and the underlying superficial lamina propria have been constructed and solved in order to ascertain the influence of structural changes associated with malignancy, oedema and inflammation on the measured electrical properties of the tissue. Sensitivity analysis of results from a composite tissue model suggests that the increase in lymphocyte density in the lamina propria associated with an inflammatory response to the infiltration of urine into the tissue may explain these unusual electrical properties.

Blood glucose control by intermittent loop closure in the basal mode: computer simulation studies with a diabetic model.

A semiclosed loop, bedside insulin infusion system using a simple basal infusion algorithm consisting of a linear transition between two insulin delivery rates as blood glucose (BG) increases has been developed. A theoretical study using computer simulation has now been undertaken to examine the effect of BG sampling frequency and algorithm parameters on BG control. A model for BG control by exogenous insulin in the individual with diabetes was developed from a model for healthy subjects and from clinical data in the literature. Results of computer simulation using this model showed a decrease in BG stability as the sampling interval increased from 1 to 4 h. Simulations also showed a decrease in BG stability as the sensitivity of the control algorithm increased. Choice of an appropriate basal control algorithm involved a compromise between stability, sampling interval, and metabolic control. We conclude that satisfactory metabolic control can be obtained using intermittent BG sampling in the basal state; sampling at intervals of 3 h combined with a basal control algorithm whereby insulin delivery rate increases linearly from 0.5 to 2.5 U/h over the BG range 2-12 mmol/L appears suitable for most diabetic persons. Three-hour sampling offers a good compromise between degree of metabolic control and clinical effort involved.

Development and validation of computational models of cellular interaction.

In this paper we take the view that computational models of biological systems should satisfy two conditions - they should be able to predict function at a systems biology level, and robust techniques of validation against biological models must be available. A modelling paradigm for developing a predictive computational model of cellular interaction is described, and methods of providing robust validation against biological models are explored, followed by a consideration of software issues.

Virtual biopsies in Barrett's esophagus using an impedance probe.

Preliminary results of electrical impedance measurements in squamous and columnar epithelia in rat and human tissues are presented. The aim of this work is to show the possibility of differentiating these two types of epithelia in terms of their electrical characteristics. For the measurements, we employed a 1.95-m-long, 3.2-mm-diameter, four-electrode probe designed to be used transendoscopically in the diagnosis of Barrett's esophagus (BE). BE is a condition in which the normal squamous epithelium of the esophagus is replaced by columnar epithelium of the intestinal type. This metaplasia is considered as a premalignant condition that puts patients at a 30-125-fold risk of developing adenocarcinoma of the esophagus. The diagnosis and surveillance of BE involve taking multiple biopsies, an expensive and time-consuming procedure. This study constitutes the first stage in the replacement of tissue biopsy by "virtual biopsies".

Agent-based computational modeling of wounded epithelial cell monolayers.

Computational modeling of biological systems, or in silico biology, is an emerging tool for understanding structure and order in biological tissues. Computational models of the behavior of epithelial cells in monolayer cell culture have been developed and used to predict the healing characteristics of scratch wounds made to urothelial cell cultures maintained in low- and physiological [Ca2+] environments. Both computational models and in vitro experiments demonstrated that in low exogenous [Ca2+], the closure of 500-microm scratch wounds was achieved primarily by cell migration into the denuded area. The wound healing rate in low (0.09 mM) [Ca2+] was approximately twice as rapid as in physiological (2 mM) [Ca2+]. Computational modeling predicted that in cell cultures that are actively proliferating, no increase in the fraction of cells in the S-phase would be expected, and this conclusion was supported experimentally in vitro by bromodeoxyuridine incorporation assay. We have demonstrated that a simple rule-based model of cell behavior, incorporating rules relating to contact inhibition of proliferation and migration, is sufficient to qualitatively predict the calcium-dependent pattern of wound closure observed in vitro. Differences between the in vitro and in silico models suggest a role for wound-induced signaling events in urothelial cell cultures.

Electrical impedance spectroscopy (EIS) in the urinary bladder: the effect of inflammation and edema on identification of malignancy.

Previous studies have shown that tetrapolar electrical impedance spectroscopy measurements can identify cervical interstitial neoplasia with the same sensitivity and specificity as cervical smears. In the urinary bladder, the same technique yields significant differences (p < 0.05 at seven frequencies between 9.6 and 614 kHz) between normal and malignant urothelium, but is unable to classify individual measurements. Detailed histological examination demonstrates that inflammation and edema--both of which are common in abnormal urothelium--alter the impedance spectrum significantly in opposing directions. Consideration of morphological changes in abnormal urothelium suggests alternative measurement strategies.

A two-dimensional Kolmorogov-Smirnov test for binned data.

A new two-dimensional version of the Kolmorogov-Smirnov test has been developed for the testing of data which has been placed in a two-dimensional array of bins. Monte Carlo testing has been used to derive the test statistics for 8 x 8, 16 x 16 and 32 x 32 arrays. The predicted probabilities for non-square arrays and arrays with intermediate numbers of elements are compared to the probabilities derived from Monte Carlo testing. The new test is applied to two-dimensional histograms of movement vectors derived from electrical impedance images of the stomach.

The physics of red cell sedimentation.

Blood was withdrawn from the ante-cubital vein of five normal volunteers and the erythrocytes were washed and resuspended in isotonic saline. The rate of descent of the erythrocyte-saline boundary was measured every hour for a period of at least 4 h for different suspended cell concentrations. The velocity of this boundary remained constant during the period of observation for a specific concentration. There was, however, an experimental variation of sedimentation rate with red cell concentration. Extrapolation to zero concentration allowed the 'Svedberg constant' for normal red cells to be evaluated. A continuum physics theory was used to explain the observations. The sedimentation rate was also measured in anticoagulated whole blood. The rates of descent in whole blood and the red cell-saline suspension were significantly different. Furthermore, the erythrocyte boundary in whole blood displayed both acceleration and deceleration during the period of observation. These experiments support the concept, proposed in an earlier theoretical study, of an internal energy dissipative mechanism which occurs in heterogeneous fluids (such as blood) but not in homogeneous fluids (such as water).

Propranolol elimination as described by the venous equilibrium model using flow perturbations in the isolated perfused rat liver.

The kinetics of hepatic elimination of the high-clearance drug propranolol has been interpreted in a previous study from our laboratory, in which propranolol protein binding was varied, to conform to the venous equilibrium model. In another study by a different group, in which perfusate flow was varied, propranolol kinetics was interpreted to conform to the sinusoidal model. In the present study, we investigated the possibility that this discrepancy is due to the use of the two different discriminants, flow and protein binding, in the two studies. In eight livers, perfused in a recirculating design, steady-state elimination of propranolol (infused at a rate of 22.8 micrograms/min) was examined at perfusate flow rates of 16 and 32 mL/min. Hepatic outflow concentration was independent of perfusate flow rate, while the logarithmic average concentration was significantly lower at the higher flow. These data conform to the venous equilibrium model and are not consistent with the sinusoidal model. This shows that the outcome of these modeling experiments does not depend on the experimental approach used, and reaffirms that the venous equilibrium model is appropriate for propranolol under the conditions studied.

The epitheliome: agent-based modelling of the social behaviour of cells.

We have developed a new computational modelling paradigm for predicting the emergent behaviour resulting from the interaction of cells in epithelial tissue. As proof-of-concept, an agent-based model, in which there is a one-to-one correspondence between biological cells and software agents, has been coupled to a simple physical model. Behaviour of the computational model is compared with the growth characteristics of epithelial cells in monolayer culture, using growth media with low and physiological calcium concentrations. Results show a qualitative fit between the growth characteristics produced by the simulation and the in vitro cell models.

Data processing techniques for serial EIT spectroscopy images: a review of some preliminary results.

Mutlifrequency EIT imaging should allow specific organs within the body to be identified by their impedance spectrum, and the use of parametric imaging should lead to a much greater freedom from movement artefacts. This will make EIT more attractive as a monitoring technique, but the data rate will require automated processing of the images. The application of dynamic regions of interest, generated on a frame by frame basis, is described, with examples from the imaging of neonatal lungs and adult stomach. The lung can be objectively identified on a single frame from the fRSC, SC and RC images, but the stomach could only be identified on the dynamic images.

Transport of gastric contents.

A planar array of electrodes has been used to provide a longitudinal section of the stomach. Impedance changes at the gastric frequency of 0.05 Hz can be detected. The changes are mainly located around the periphery of the stomach image, suggesting that they are the result of movement of the stomach wall. The generation of a vector histogram of wall movement gives a non-invasive method which appears to quantify the peristaltic waves which produce transport in the stomach.

Movement artefact rejection in impedance pneumography using six strategically placed electrodes.

In this paper, we have proposed a technique for reducing movement artefacts in impedance pneumography by placing six electrodes at appropriate locations and suitably combining the measurements obtained. The strategy for electrode placement was based on the observation that the electrodes appeared to slide over the rib cage along with the skin, during movement. A volume conductor model of the thoracic cavity was developed and movement artefacts were simulated by shifting the electrodes to a different location on the surface. The impedance changes due to movement in one of the measurements of a 'symmetrical pair' were 180 degrees out of phase with respect to those observed in the other measurement of that pair. However, the impedance changes due to breathing were in phase in both these measurements. Thus, it was possible to reduce movement artefacts by taking a mean of these measurements without affecting the breathing related changes. The six electrodes could be configured into two such symmetrical pairs. The same observation was made in experimental data recorded from human subjects. This indicated that movement artefacts were caused by sliding of electrodes along with the skin and could be reduced by using the six-electrode configuration.

Cole equation modelling to measurements made using an impulse driven transfer impedance system.

Electrical impedance measurements are used to obtain information about a subject, tissue sample or tissue model under test. There are several ways of obtaining these impedance data and thereafter analysing the data to obtain relevant parameters. This paper shows how a completely isolated drive and receive system using current pulses, as opposed to sine waves, achieves good fitted results with resistor-capacitor Cole phantoms.

Detecting oesophageal-related changes using electrical impedance tomography.

Preliminary work has been carried out using the Sheffield mark II real time EIT system, looking for changes in conductivity which occur in the stomach and oesophagus following a swallow of a small volume of either a conducting or a non-conducting liquid. This has been done using three different configurations: a conventional transverse array placed around the thorax, a rosette array on the abdomen, and a rosette array placed on the back. Results show a significant difference between the two liquids, which can be detected equally well by the three electrode configurations.

Multifrequency and parametric EIT images of neonatal lungs.

The aims of the study were to investigate the problems involved in making multifrequency EIT measurements on neonates and to compare the images obtained with the results from a group of normal adults. The Sheffield electrical impedance tomographic spectroscopy (EITS) system acquires multifrequency data using a set of eight drive and eight receive electrodes. EITS measurements were made on an inhomogeneous group of 10 neonates admitted to the special care baby unit for observation and feeding. R/S, characteristic frequency, RC and SC parameters were generated using the Cole equation. Comparisons of the parameters were made with data collected from normal adults in another study. We have shown that it is possible to obtain EITS parametric images of neonatal lungs and that there are some differences in Cole parameters between the adult and neonatal groups.

A study of the morphological parameters of cervical squamous epithelium.

Electrical impedance spectroscopy is a technique that has been investigated as a potential method for the diagnosis of epithelial carcinomas. Finite element modelling can provide an insight into the patterns of current flow in normal and pathological epithelium and hence aid in the process of probe design optimization. In order to develop a finite element model of the structure of normal and precancerous cervical squamous epithelium, it was first necessary to obtain the mean values and ranges of a number of morphological tissue parameters. The most important parameters in discriminating normal from neoplastic tissue were identified as being cell size and shape distribution, nuclear-to-cytoplasmic volume ratio and volume of extracellular space. A survey of the literature revealed an absence of reliable quantitative data for these parameters. We therefore present the results of our own basic image analysis on normal and pathological tissue sections, which we hope will be of use to other workers wishing to model cervical squamous epithelium, or other similar tissue structures.