Table of contents

A scientific journal serves two primary roles: the presentation of timely, high-quality research and the archiving of such material for future reference. Secondary roles, fulfilled by many journals, are to provide well-documented tutorial reviews of currently important research in the journal's area of emphasis as well as to provide a venue for limited scientific discussion of essential issues in the field. As I assume the position as Honorary Editor of Physiological Measurement, I realize how important these roles are in the success of a journal and the great responsibility that the editor and the publisher have in guiding the publication to fulfil them. It is an honour and a challenge to take on this responsibility. I am pleased to follow the strong example set by my predecessor, Stuart Meldrum, to work with the publisher, Jane Roscoe, and the rest of the staff at Institutue of Physics Publishing (IOPP), and to receive advice and assistance from an outstanding international Editorial Board.

Dr Meldrum and the staff at IoPP have steered the journal on a steady course of increasing impact factor, decreasing receipt-to-publication processing times, and increasing subscriptions. Dr Roscoe and the IoPP staff have established a state-of-the-art electronic manuscript submission, review and electronic publishing capability for the journal. In 2001 there were over 53,000 hits on Physiological Measurement's home page with over 8,500 downloads of manuscripts; both values showing significant increases from the previous year.

Our Editorial Board has representatives from Europe, North America, Australia, and Asia, and one of my goals as editor will be to further increase the international presence on this Board. In this regard, I am pleased to announce that Professor Idagene Cestari from Sao Paulo, Brazil and Dr Jean-Pierre Morucci from Toulouse, France, who is past president of the International Federation of Medical and Biological Engineering (IFMBE), have joined the Board this year.

The most important groups in the publication process are you the authors, reviewers and readers. For a journal to be successful, it must meet your needs and do so in a timely, efficient way. We, who are involved in the journal's production, have a great responsibility to you to provide an effective means to disseminate your work and to meet your needs to know the state-of-the-art in physiological measurements. You also have a responsibility to us to submit appropriate, high-quality manuscripts for consideration and to serve as peer reviewers in an expeditious way when called upon to do so. A quality publication must have capable, fair and considerate reviewers as well as outstanding authors and manuscripts.

Thus, we are all in this together: authors, reviewers, readers, editor, and publisher. Through our combined efforts we can continue along the path established by my predecessors. We can continue to bring Physiological Measurement to the forefront as the première journal in which to publish manuscripts, topical reviews, notes and brief correspondence on the theory, technology, techniques, and understanding on how we can better measure and comprehend physiological systems.

1. Introduction

This special (part) issue of Physiological Measurement is the third annual edition produced in relation to conferences held each year in London under the auspices of the EPSRC Engineering Network in Biomedical EIT. In this issue (from page 95) there are sixteen papers which reflect the healthy state of interest and activity in biomedical EIT; these represent a selection of the presentations at the last annual network conference held in London in April 2001. There has again been good progress in all fields covered - algorithms, hardware, theory and clinical applications. Promising new areas include magnetic induction tomography (MIT), the now commercial use of impedance mapping for breast cancer diagnosis, the potential use of bioimpedance to diagnose cervical cancer, and EIT of brain function. All are reflected in papers in this issue.

However, EIT has not yet made the transition from an exciting medical physics discipline into widespread routine clinical use. The engineering network was primarily created in 1998 to try and further this by co-ordinating a multi-centre clinical trial. As the official funding for the network finished in September 2001, I have included in the next paragraphs a summary of the achievements as I see them.

2. EPSRC Engineering Network in Biomedical EIT (1998-2001)

2.1 Application for a multi-centre clinical trial of electrical impedance tomography

Shortly after the establishment of the engineering network in 1998, a workshop was held for interested parties. The leaders of approximately eight groups all met to establish a steering committee for a large-scale clinical application. No international agreement had previously been achieved on this issue, and there was concern lest each group might wish to use its own EIT system, so that a common platform could not be agreed.

In the event, matters proceeded very smoothly. The following plan was commonly agreed: a common hardware and software platform would be used by each of the participating groups. This was agreed to be the latest EIT system from Professor Brian Brown's group at Sheffield, termed the `Sheffield Mark 3'. This system was based on systems which were by far the most widely used in published clinical studies and included the new development of multi-frequency recording. In contrast, each group was free to pursue clinical applications in their own sphere of interest. This included recordings in medical conditions in heart and chest medicine, gastric emptying, breast imaging, and imaging of brain function. A working plan was put forward, and this was put to all attendees at the next international conference which was held in London in April 2000.

The plan was commonly agreed at this conference and ten groups agreed to participate. Enquiries were made to suitable grant-giving bodies in order to see which of them would be suitable for this application. Of the UK national bodies, the only one which would permit an application of this type was the Medical Research Council. Unfortunately, the application was rejected on strategic, not scientific, grounds; the referees' scientific reports were attached and were strongly supportive. The difficulty with this sort of application is that it is not strictly a clinical trial, in the sense that it is not a trial of therapy. It was for a relatively large sum - about £1 million - and so would have been too large for a project grant application to one of the research councils. At least in the UK, there does not appear to be suitable funding source for this sort of project; it does not qualify as a clinical trial, and is too early to be funded as a trial by a commercial concern.

Agreement had thus been obtained between participating groups, and the matter was raised at subsequent conferences. In 2001, a further application was made to the basic technology call from the combined research councils, but an outline application for this was unfortunately also rejected.

Further discussion at subsequent conferences led to the proposition that the chance of success may be greater if the application was for a single clinical field, rather than across several different ones, as in the original proposal. An application for four collaborating groups in EIT of chest function, organized by Anca Boonstra (Amsterdam), is currently being considered by the European Community.

The Engineering Network has therefore been successful in providing a forum for hitherto disparate groups to agree on a common approach and platform, and a collaboration in the field of assessing EIT for imaging in chest and heart function is currently still being considered. It is unfortunate that the full applications to the MRC or research councils were not successful, but the field is still active and healthy, as individual groups are pursuing EIT clinical trials in the fields of chest, heart, breast, and brain imaging. As the network is now self-sustaining (see below), international effort into clinical trials can still continue in a co-ordinated way.

2.2 Development of a common software platform for the IT community

A direct result of the collaborations developed at the international conferences has been the development of a common software suite for reconstruction of EIT images. There are numerous different possible approaches for mathematical reconstruction of EIT images. Until now, individual groups have produced their own algorithms and, as a result, there has been considerable duplication of resources. With the help of a network funded workshop in 2000, Bill Lionheart from UMIST, in collaboration with several different members of the network, has produced a common software platform for reconstruction of the EIT images (EIDORS). This is written in Matlab so that it can be implemented on different platforms and is a generally available resource for any researcher in EIT.

2.3. Website and EIT discussion list

At present, there is no specific society for electrical impedance tomography. However, funded by the network, we have set up a website which serves the national and international EIT community. It has been used as the repository of information about the network, publications, and other relevant information. We have also set up a discussion mailing list which provides a common informal forum for discussion and exchange of information.

3. Future perspectives for biomedical EIT

The biomedical EIT research community has been fortunate in that there have been annual conferences and special issues in most years since 1987, so there is a forum for exchanging ideas and making a common effort. The EPSRC Engineering Network has been particularly valuable over the past three years in allowing us to continue this tradition and concentrate on establishing clinical trials. There is currently no formal mechanism to continue this, but Richard Bayford and I, here at University College London, are happy to continue to maintain our virtual community whether further explicit funding becomes available or not. The website and discussion list will continue (www.eit.org.uk), and we shall continue to organize annual conferences, probably in London. In this year (2002), David Isaacson and Jennifer Mueller have organized a conference to be held in Colorado, USA, which will take its place (www.eitworkshop.org). We hope, therefore, that the tradition will continue.

The state of EIT research is still healthy. There are about 20 or 30 groups worldwide who are actively performing research, and I think it is still seen as an exciting area of medical physics. In the past year or two, there has been a breakthrough into clinical practice, with the granting of FDA approval for the use of impedance scanning hardware on a commercial basis for the adjuvant diagnosis of breast cancer. However, a major drawback is that it is not yet accepted for routine clinical use in any other area, in spite of the existence of a score or so of good pilot studies. As mentioned above, I think a major element of this is that the clinical community needs to be convinced if its utility by convincing large scale clinical trials; at present, the great majority of researchers in the field are physicists or engineers, and doctors need to be involved to see the benefits of the technology.

In terms of technical developments, there appear to be two main streams. One stream is to concentrate on tissue characterization by multifrequency recording, and to use the imaging ability of EIT to delineate a small number of regions of interest. This has been the approach of the Sheffield group, whose Mark 3.5 system usually has only eight channels. My personal view is to press on with the development of the imaging capability of EIT; our new system for brain imaging has 64 electrodes and records several hundred different electrode combinations. This approach certainly works reasonably well in tanks, but there are practical limitations in human recordings. Other groups producing new systems appear to lie between these two viewpoints.

It is not entirely clear where the bottleneck in EIT image and data quality lies; my guess is that it is principally in the errors produced by the interaction of skin impedance with instrumentation. These errors are then magnified by the ill-conditioned inverse solution in the reconstruction algorithms. Although there have been several pilot publications on the use of active electrodes to reduce errors, a practical system which employs these may well be the next major advance in our field. At the same time, there is considerable interest in magnetic induction tomography, which avoids the skin interface problem, and I look forward to seeing the first clinical trials with this novel technology.

In conclusion, the past three years have seen a healthy, steady advance in the field. We still need to break into widespread clinical acceptance, and effort is continuing actively into the clinical trials and pilot studies which will achieve this. Continuation of a virtual research community over the web and with annual conferences will allow research groups to co-ordinate and collaborate in achieving this. Technical advances may allow us to obtain more accurate tissue characterization and image quality, and this will undoubtedly help to advance clinical acceptance.

This review first gives an overview on the concept of fractal geometry with definitions and explanations of the most fundamental properties of fractal structures and processes like self-similarity, power law scaling relationship, scale invariance, scaling range and fractal dimensions. Having laid down the grounds of the basics in terminology and mathematical formalism, the authors systematically introduce the concept and methods of monofractal time series analysis. They argue that fractal time series analysis cannot be done in a conscious, reliable manner without having a model capable of capturing the essential features of physiological signals with regard to their fractal analysis. They advocate the use of a simple, yet adequate, dichotomous model of fractional Gaussian noise (fGn) and fractional Brownian motion (fBm). They demonstrate the importance of incorporating a step of signal classification according to the fGn/fBm model prior to fractal analysis by showing that missing out on signal class can result in completely meaningless fractal estimates. Limitation and precision of various fractal tools are thoroughly described and discussed using results of numerical experiments on ideal monofractal signals. Steps of a reliable fractal analysis are explained. Finally, the main applications of fractal time series analysis in biomedical research are reviewed and critically evaluated.

Currently retinal imaging is performed with the fundus camera. This has a number of limitations, in particular the high level of illuminations required for imaging. The scanning laser ophthalmoscope (SLO) has been proposed as an alternative imaging device but to date one of its main limitations has been that it gives only monochromatic images. In this paper we describe an SLO which uses low power red, green and blue lasers to image the human fundus. Using three lasers simultaneously to produce a colour image will increase the fundus exposure by a factor of three. To overcome this problem, a technique has been developed for multiplexing the lasers so that each point on the retina is imaged by the three lasers pulsed rapidly in sequence. The total exposure is thus kept to the same level as for a single laser and total imaging time is not increased. An example is shown of the image from a patient with diabetic retinopathy.

Cardiopulmonary bypass (CPB) is a safe support system during major cardiac operations because it allows the distribution of adequately oxygenated blood to the entire body. The optimal perfusion flow to be used during hypothermic CPB remains a controversial issue. In the present study, the effect of different flows and pressures on oxygen delivery, extraction and consumption was compared. Such an effect was evaluated for 40 patients undergoing open-heart surgery using CPB (membrane oxygenator) and systemic hypothermia. The patients were randomized into two groups (20 each) according to the flow and pressure (low flow, low pressure and high flow, high pressure). The electrical impedance from 50 Hz to 10 MHz for blood samples and erythrocytes for both groups was also measured at different time intervals (before-, during and post-CPB). No significant change for low flow group was recorded (p > 0.05) while the dielectric properties of the high-flow group were greatly affected by the time of operation. Hence, low flow can be safely used in young patients without any organ impairment or carotid artery disease, while high flow should be used for patients with any organ dysfunction, at the initiation of CPB and during rewarming to compensate for the increase in oxygen consumption and the need for better perfusion pressure.

The distribution function of pulmonary transit times (fPTTs) contains information on the transit time of blood through the lungs and the dispersion in transit times. Most of the previous studies have used specific functional forms with adjustable parameters to characterize the fPTT. It is the purpose here to investigate the possibility of estimating the fPTT in a model-free way. The method employs the maximum entropy principle and is used, in particular, on cardiac positron emission tomographic (PET) studies but is believed to be more generally applicable. Using this principle in a test case, we were able to accurately identify a two-peaked transfer function, which may theoretically be seen in patients with pulmonary disease confined to one lung. Transit time values for [¹³N]-ammonia were produced by applying the algorithm to PET studies from normal volunteers.

The results of a preliminary clinical evaluation of a one-frequency electrical impedance tomography (EIT) system enabling static in vivo imaging are presented. The design of the measuring system and image reconstruction software are described. Thirty-one subjects were examined and divided into four clinical groups. The first group consisted of 22 patients with clinical diagnosis of lung cancer with tumour localization in one lung. The second group consisted of seven healthy subjects. A patient after a one-sided pneumectomy and another with one-sided emphysema diagnosis were also examined. Static EIT images of a healthy human chest and a chest with various abnormalities are given and discussed. The evaluated system distinguishably visualizes various states of lungs and thorax including lung cancer. The average static conductivity of an affected lung in the first clinical group statistically differs from the average conductivity of a healthy lung. In spite of low spatial resolution, according to preliminary results, the method can be sensitive to cancer and other lung diseases in screening investigations.

The generalizability theory, an expansion of classic true-score reliability theory, was used to investigate the generalizability of observed segmental extracellular fluid (ECF) and intracellular fluid (ICF) distribution measurements. The test instrument was a Xitron Hydra ECF/ICF bioimpedance analysis system model 4200, Xitron Technologies, San Diego, CA. Fifty American healthy men (17–72 years) and 50 American healthy women (17–76 years) volunteered as participants. Xitron continuous segmental ECF–ICF procedures for testing leg segmental data were followed for testing participants in both the standing erect and lying supine postures. A two-facet, person-by-trial, completely crossed design was used. All facets were treated as random. During a one-day session each subject was tested involving 20 trials for the standing erect posture and 20 trials for the lying supine posture. Data on each fluid measurement, each body posture and each sex group were independently analysed. The analyses revealed that the trial factor accounted for less than 0.2% of the total variance for ECF and ICF scores. ECF and ICF generalizability coefficients for the segmental method were 0.99 or greater. In comparing ECF segmental to ECF global, results showed generalizability coefficients were similar. However, ICF segmental coefficients were larger than the coefficients produced by the global method. In conclusion, the segmental method appeared more reliable than the global method, under the conditions of this study.

Assessments of dynamic cerebral autoregulation usually measure the cerebral blood flow velocity (CBFV) response to changes in arterial blood pressure (ABP). We studied the effect of substituting ABP by cerebral perfusion pressure (CPP), expressed as the difference between ABP and intracranial pressure (ICP), in estimates of dynamic autoregulation obtained by transfer function analysis. CBFV, ABP and ICP were recorded during periods of physiological stability in 30 patients with severe head injury. Transfer function analysis was performed using the following combinations of input–output variables: ABP–CBFV, CPP–CBFV and CBFV–ICP. Frequency and time-domain (step response) functions were averaged for recordings with mean ICP < 20 mmHg (group A) and mean ICP >= 20 mmHg (group B). The ABP–CBFV transfer function parameters and step response for group A were similar to previous studies in normal subjects, but group B showed deterioration of dynamic autoregulation. Radically different step responses were obtained from both groups for the CPP–CBFV transfer function and the coherence was not significantly improved. The CBFV–ICP transfer function had the highest values of coherence and indicates that changes in CBFV are the cause of spontaneous fluctuations in ICP. Furthermore, the ICP step response plateau was significantly higher for group B than for group A. An alternative calculation of the CBFV step response to changes in CPP resembled the corresponding responses for the ABP input. For spontaneous fluctuations in ABP, ICP and CBFV, it is not possible to calculate the CPP–CBFV transfer function directly due to the high positive correlation between ICP and CBFV, but an alternative estimate can be obtained by using the CBFV–ICP transfer function. The latter could also be useful as a method to assess intracranial compliance in head injury patients.

A forced periodic variation in blood pressure produces a similar variation in cerebral blood velocity. The amplitudes and phases of the pressure and velocity waveforms are indicative of the dynamic response of the cerebral autoregulation. The phase of the velocity leads the pressure; the greater the phase difference the faster the autoregulation response. Various techniques have been employed to oscillate arterial blood pressure but measurement reproducibility has been poor. The purpose of this study was to assess the reproducibility of phase measurements when sinusoidal lower body negative pressure is used to vary blood pressure.

Five healthy volunteers were assessed at two vacuum levels on each of eight visits. For each measurement a 12 s sinusoidal cycle was maintained for 5 min. The Fourier components of blood pressure and the middle cerebral artery velocity were determined at the oscillation frequency.

The phase of velocity consistently led the pressure. The mean phase difference was 42 ± 13° for the stronger vacuum and 36 ± 42° for the weaker vacuum. The variation given is the within-subjects standard deviation estimated from a one-way analysis of variance.

Sinusoidal lower body negative pressure is a useful stimulus for investigating autoregulation; it has advantages over other methods. High vacuums show good reproducibility but are too uncomfortable for patient use.

Blood pressure pulse wave velocity (PWV) is a parameter which is related to arterial distensibility. Its direct assessment, by measuring the appearance time of a pressure pulse in two sites along an artery and the distance between the two sites, is complicated and inaccurate. In the current study, pulse transit time (PTT) to the toes and fingers of 44 normotensive male subjects was measured by photoplethysmography (PPG) and ECG. The arrival time of the pulses at the toe and finger was determined from the foot of the systolic rise of the PPG signal, i.e. at end-diastolic time. Two parameters, which are related to PWV, were tested: the time delay between the ECG R-wave and the arrival time of the pulses at the toe (E-T PTT), and the difference in the transit time of the blood pressure pulses between the toe and finger (T-F PTTD). E-T PTT and T-F PTTD decreased as functions of the subject's age and systolic blood pressure (SBP), but their dependence on the diastolic blood pressure (DBP) was not statistically significant. The decrease of the PTT parameters with age is attributed to the direct structural decrease of the arterial compliance with age and not to functional effects associated with the increase of the blood pressure with age, since the PTT parameters did not depend on DBP though the measurements were performed at end-diastole.

Non-linear electrical impedance tomography reconstruction algorithms usually employ the Newton–Raphson iteration scheme to image the conductivity distribution inside the body. For complex 3D problems, the application of this method is not feasible any more due to the large matrices involved and their high storage requirements. In this paper we demonstrate the suitability of an alternative conjugate gradient reconstruction algorithm for 3D tomographic imaging incorporating adaptive mesh refinement and requiring less storage space than the Newton–Raphson scheme. We compare the reconstruction efficiency of both algorithms for a simple 3D head model. The results show that an increase in speed of about 30% is achievable with the conjugate gradient-based method without loss of accuracy.

Electrical impedance tomography (EIT) may be used to image brain function, but an important consideration is the effect of the highly resistive skull and other extracerebral layers on the flow of injected current. We describe a new reconstruction algorithm, based on a forward solution which models the head as four concentric, spherical shells, with conductivities of the brain, cerebrospinal fluid, skull and scalp. The model predicted that the mean current travelling in the brain in the diametric plane for current injection from polar electrodes was 5.6 times less than if the head was modelled as a homogeneous sphere; this suggests that an algorithm based on this should be more accurate than one based on a homogeneous sphere model. In images reconstructed from computer-simulated data or data from a realistic saline-filled tank containing a real skull, a Perspex rod was localized to within 17% or 20% of the tank diameter of its true position, respectively. Contrary to expectation, the tank images were less accurate than those obtained with a reconstruction algorithm based on a homogeneous sphere. It is not yet clear if the theoretical advantages of this algorithm will yield practical advantages for head EIT imaging; it may be necessary to proceed to more complex algorithms based on numerical models which incorporate realistic head geometry. If so, this analytical forward model and algorithm may be used to validate numerical solutions.

In non-linear electrical impedance tomography the goodness of fit of the trial images is assessed by the well-established statistical χ² criterion applied to the measured and predicted datasets. Further selection from the range of images that fit the data is effected by imposing an explicit constraint on the form of the image, such as the minimization of the image gradients. In particular, the logarithm of the image gradients is chosen so that conductive and resistive deviations are treated in the same way. In this paper we introduce the idea of adaptive mesh refinement to the 2D problem so that the local scale of the mesh is always matched to the scale of the image structures. This improves the reconstruction resolution so that the image constraint adopted dominates and is not perturbed by the mesh discretization. The avoidance of unnecessary mesh elements optimizes the speed of reconstruction without degrading the resulting images. Starting with a mesh scale length of the order of the electrode separation it is shown that, for data obtained at presently achievable signal-to-noise ratios of 60 to 80 dB, one or two refinement stages are sufficient to generate high quality images.

In electrical impedance tomography, many factors affect the image reconstruction results. Among them are the number of electrodes (NOE) and the number of conductivity basis functions (NOCBF) for image reconstruction. The NOCBF generally reflects the density of the mesh with which images are reconstructed. How and to what extent do these factors affect the image reconstruction and corresponding images? In this area detailed analysis is still lacking. This study aims to address the above question.

In this study, image reconstruction and its ill-posed condition were analysed by singular value decomposition (SVD) and spectral expansion theory with different configurations of NOE and NOCBF. The results in this study indicate that for a circular 2D plane object with electrodes evenly located around the boundary: (1) Under certain conditions, increasing the NOE enables us to improve the ill-posed condition in image reconstruction and hence improve the image quality. Generally more improvement is expected near the image periphery than in the image centre. (2) Increasing the NOCBF generally worsens the ill-posed condition. But it enables the solution to be sought in a finer subspace and may be able to improve the image quality on the periphery, while generally the result in image centre depends more on the prior information incorporated in the regularization.

In electrical impedance tomography, the shape of the object being imaged (such as the human thorax) is often complex. For this reason, numerical techniques, such as finite element method, are often used for solving the forward problem in 3D rather than analytical solutions which can only model simple geometrical shapes. However, an analytical solution to the 3D forward problem can often be useful. This paper will present an analytical solution to the forward problem for an elliptical cylinder whose eccentricity can be easily modified to approximate the shape of the human thorax.

The UCLH Mark 1b is a portable EIT system that can address up to 64 electrodes, which has been designed for imaging brain function with scalp electrodes. It employs a single impedance-measuring circuit and multiplexer so that electrode combinations may be addressed flexibly using software. It operates in the relatively low frequency band between 225 Hz and 77 kHz, as lower frequencies produce larger changes during brain activity, and has a videocassette-sized headbox on a lead 10 m long, connected to a base box the size of a video recorder, and notebook PC, so that recordings may be made in ambulant subjects.

Its performance was assessed using a resistor–capacitor network, and two saline-filled tanks—a cylindrical Perspex one and a latex one which contained a human skull. System signal-to-noise ratio was better than 50 dB and the maximum reciprocity error less than 10% for most frequencies. The CMMR was better than 80 dB at 38 kHz and a sponge, 20 mm across, which caused a local 12% impedance increase, was correctly localized in images. This suggests that the system has adequate performance to image impedance changes of 5–50% known to occur in the brain during normal activity, epilepsy or stroke; clinical trials to image these conditions are in progress.

The electrical properties of cervical squamous epithelium have been modelled in the frequency range 100 Hz to 10 MHz. The hierarchical modelling process comprises a cellular level stage, which includes detailed models of cells typical of different depths within the epithelium and a tissue model, which utilizes electrical properties obtained from the cellular models. The fit between the modelled and measured impedance spectra and the distribution of current with depth depends on the macroscopic model structure. Both the properties of the basement membrane and the presence of a surface mucus layer are shown to have a significant effect. The best fit with measured data is obtained when a 10 μm thick, high-conductivity surface layer is included in the tissue model.

The literature concerning measurement of spatial resolution in electrical impedance tomography (EIT) is vague. Different groups often use their own method or a modified version of a better known method, thus hindering a generalized resolution measurement which could be useful for gauging the performance of one system against another. Measurement of spatial resolution in EIT is further complicated by its spatial variant nature and hence cannot be expressed simply with a single parameter as it can be in other imaging modalities (such as nuclear medicine or MRI for example). If the performance of each acquisition and image reconstruction system in EIT is to be compared objectively then there needs to be a common standard. In this paper the results of different methods for calculating spatial resolution are compared and an improved method is proposed which aims to fulfil this role.

One of the problems facing anyone attempting the investigation of dielectric properties of living tissue is the presence of skin, which screens all that lies under it from direct measurement. Thus, in non-invasive breast examination using transimpedance measurements, skin parameters heavily influence the results, specifically at low (less than 10 kHz) frequencies. In this paper a method for overcoming this difficulty by using multi-frequency measurements obtained from a surface current distribution over a flat probe is described. By using the variation in the shape of the real and imaginary parts of the surface current density at different frequencies, the original dielectric values of the skin and the underlying tissue can be obtained, based on the assumption of the existence of a two-layer geometry, with the upper (skin) layer much thinner than the lower (tissue) layer. The results obtained can be used in the diagnosis of breast cancer using existing transimpedance measurement devices.

This paper reports a preliminary finding associated with an investigation of how tissues respond to mechanical stress. The stress distribution within the tissue may be the result of normal function, for example, joint forces, or it may result from interventions such as tissue suturing during or after surgery. We sought to combine electrical and mechanical computational models in order to better understand the interaction between the two. For example, if mechanical stress is applied to tissue this may change the cell arrangements within the tissue matrix and hence change the electrical properties. If this interaction could be determined, then it should be possible to use electrical impedance tomography measurements to identify stress patterns in tissues.

Measurements of resistivity changes have been made in conductive silicone rubber sheets when subject to a uniaxial stress of up to 10%. Relatively large changes in resistivity are produced (up to 200%). These changes are far larger than those predicted arising from topological changes alone. It is suggested that under stress the conductive islands of carbon within the silicone rubber sheet undergo a reversible disassociation from their neighbours and that the material's electrical properties change under load. If similar stress–resistivity relationships occur within biological materials it may be possible to recover the stress fields within tissues from transfer impedance measurements and thereby predict if actions such as inappropriate suture tension will compromise tissue viability.

High-frequency (3–30 MHz) operation of MIT systems offers advantages in terms of the larger induced signal amplitudes compared to systems operating in the low- or medium-frequency ranges. Signal distribution at HF, however, presents difficulties, in particular with isolation and phase stability. It is therefore valuable to translate received signals to a lower frequency range through heterodyne downconversion, a process in which relative signal amplitude and phase information is in theory retained. Measurement of signal amplitude and phase is also simplified at lower frequencies.

The paper presents details of measurements on a direct phase measurement system utilizing heterodyne downconversion and compares the relative performance of three circuit configurations. The 100-sample average precision of a circuit suitable for use as a receiver within an MIT system was 0.008° for input amplitude −21 dBV. As the input amplitude was reduced from −21 to −72 dBV variation in the measured phase offset was observed, with the offset varying by 1.8°. The precision of the circuit deteriorated with decreasing input amplitude, but was found to provide a 100-sample average precision of <0.022° down to an input amplitude of −60 dBV. The characteristics of phase noise within the system are discussed.

Magnetic induction tomography (MIT) is a contactless method for mapping the electrical conductivity of tissue by measuring the perturbation of an alternating magnetic field with appropriate receiver coils. Reconstruction algorithms so far suggested for biomedical applications are based on weighted backprojection, hence requiring tube-shaped zones of sensitivity between excitation coils and receiving coils, the sensitivity being essentially zero outside this 'projection beam'. This condition is met for conducting perturbations in empty space and for some special configurations of insulators in saline. In biological structures, however, perturbations with low conductivity contrast are embedded into a bulk conductor. The respective sensitivity distribution was investigated and quantified theoretically and experimentally by displacing a conducting (agar, 8 S m⁻¹) and an insulating sphere within a saline tank (4 S m⁻¹). In contrast to the case in the empty space the sensitivity is not confined to a tube but even increases outside the 'projection beam'. The difference can be explained by the interaction of bulk currents with the perturbing object. This effect invalidates backprojection and hence the solution of the complete inverse eddy-current problem is suggested.

Electrical impedance images were made using the ACT 3 instrument, which applies currents simultaneously to 32 electrodes and measures the resulting voltages on those same electrodes. A reconstruction algorithm was written for a three-dimensional cylinder having electrodes in two or four layers, using current patterns that pass current among different planes of electrodes, as well as within each plane. We have previously reported useful vertical resolution by the use of added layers of electrodes. The aim of the present study was to demonstrate that physiologically useful information can be obtained by examining cephalo-caudal differences in three-dimensional images. Phasic changes throughout the cardiac cycle are seen to be markedly different at the heart compared to lung region, both above and beside it.

We formed hydrogel electrodes each 3 cm tall and 7 cm wide and applied them to the thorax of an upright human subject in four horizontal rows; each row contained eight electrodes. During breath-holding, cardiac activity was seen in all layers. With systole, conductivity in the anterior of the lowest layers decreased, but not in the upper layer. In the upper layers, conductivity increased with systole in many regions. These observations are consistent with the opposite changes in blood volume of the heart and lungs and the locations of these organs. This paper demonstrates the feasibility of producing and displaying physiologically interpretable three-dimensional images of the chest in real time.

We describe a fully automatable quantification process for the assessment of unilateral pulmonary function (UPF) by means of EIT and propose a measurement protocol for its clinical implementation. Measurements were performed at the fourth and sixth intercostal levels on a first group of ten healthy subjects (5M, 5F, ages 26–48 years) to define the proper protocol by evaluating the most common postures and ventilation modes. Several off-line processing tools were also evaluated, including the use of digital filters to extract the respiratory components from EIT time series. Comparative measures were then carried out on a second group consisting of five pre-operatory patients with lung cancer (4M, 1F, ages 25–77 years) scheduled for radionuclide scanning. Results show that measurements were best performed with the subject sitting down, holding his arms up and breathing spontaneously. As regards data processing, it is best to extract Fourier respiratory components. The mean of the healthy subject group leads to a left–right division of lung ventilation consistent with literature values (47% left lung, 53% right lung). The comparative study indicates a good correlation (r = 0.96) between the two techniques, with a mean difference of (−0.4 ± 5.4)%, suggesting that the elimination of cardiac components from the thoracic transimpedance signal leads to a better estimation of UPF.

Use of electrical impedance spectroscopy (EIS) to image the breasts of women with both normal and abnormal conditions requires the ability to deliver a consistent and repeatable exam. To investigate the degree to which our current imaging system can meet this requirement we conducted an initial study of exam consistency. The trial involved the imaging of 25 breasts stratified into four separate substudies with increasing levels of electrode placement uncertainty. The degree of complexity ranged from single-placement single-session imaging to multiple-placement single-session imaging to multiple-placement multiple-session imaging. Both visual analysis and quantitative comparisons using mean squared difference (MSD) measures between pairs of permittivity and conductivity images were performed. A new breast interface with the improved vertical and radial electrode array positioning capability required to complete this study is described. Not surprisingly, the results show a dominant trend of increased image variability with increased electrode placement uncertainty. Importantly, quantitative levels of image consistency are reported through MSD analysis. On average across all frequencies analysed, MSDs for single placements are well below 1%, near 2–3% for repositioned breasts during the same session and approximately 15% for re-examined breasts in multiple sessions conducted over time. Overall, these results suggest that EIS breast exams are consistent provided the electrode placement is well controlled, typically with better than 1 cm accuracy.

Electrical impedance tomography (EIT) has been proposed as a method to monitor dynamic changes in the pulmonary vascular bed. In this study we examined the validity of EIT in the measurement of pulmonary vasodilatation in eight patients with primary and secondary pulmonary hypertension when given the vasodilating agent epoprostenol (Flolan®). Therefore, catheterization of the pulmonary artery was performed in the ICU and the cardiac output was measured by means of the Fick method. The pulmonary vascular resistance (PVR) and mean pulmonary arterial pressure (mPAP) were determined. Epoprostenol was given in increasing doses to test reversibility of pulmonary hypertension. The maximum test dose was 12 ng kg⁻¹ min⁻¹. During each step simultaneous EIT (DAS-01 P Portable Data Acquisition System, Sheffield, England) measurements were performed with the 16 electrodes equidistantly positioned in the third intercostal space. The maximal systolic impedance change, relative to end-diastole, ΔZ_perf, was chosen as a measure of pulmonary perfusion. The impedance change between baseline and highest tolerable epoprostenol concentration was compared with the change in PVR.

The mean PVR (dyn s/cm⁵) decreased from 636 (±399) to 366 (±242): p < 0.01. ΔZ_perf (in arbitrary units) for the whole patient group increased from 901 (±295) × 10⁻³ to 1082 (±472) × 10⁻³ (p < 0.05). Only one patient showed a reduction in pulmonary artery pressure >20%, which is defined as significant vasodilatation. A strong relationship was found between the impedance changes and the change in PVR and mPAP in the patient with a significant vasodilatation on epoprostenol. From these results we conclude that EIT is a reliable method to measure blood volume changes due to pharmacologically induced vasodilatation in the pulmonary bed.

ed M Hennerici and S Meairs Publisher: Cambridge: Cambridge University Press (2001) 428pp, price: £140.00, ISBN: 0 521 63223 4

This book reminded me of a large, old-fashioned family Christmas concert. The contributors (there are 56 of them) surely know one another very well. Each has his individual talent. There is a slightly competitive frisson. Each performance is well rehearsed and each party-piece is presented with polish and verve. The gentlemen perform, and the ladies listen perhaps, there being none amongst the authors. Excellent individually, the tension and discord only appears when taken as a whole - in the community singing as it were. In the presence of such individual talent the editor-conductors find that they cannot hold the choir in unison, and accept an untidy unity of rhythm within the separate voices.

The topic gives the book its unity. It is a wide-ranging exploration of the application of ultrasound to cerebrovascular physiology and pathology. Such is the dedicated enthusiasm of some authors to demonstrate their individual knowledge that on several occasions the text spills outside these limits, for example into discussions of lower limb vascular dynamics and cardiovascular pathology, or into examples of carotid angiography. These excursions serve well to place the majority of the text in context. This is divided into three sections. The first hundred or so pages are devoted to the physics and technology of ultrasound together with some chapters on haemodynamics. It is here that the reader first realizes the difficulties faced by the editors. There is much overlap of content, and the reader would find it necessary to read three separate chapters to gain an overview of the principles of Doppler techniques. Without dwelling on this point, it must be said that there is a fine line between redundant repetition, and the need to present knowledge from different perspectives in a multi-author book, and on several occasions the text could have been easily trimmed with no loss of content. The second and largest section, seventeen chapters extending over 200 pages, is headed Clinical Cerebrovascular Ultrasound. I think this heading is a little unfair, perhaps leading the reader to expect only a tutorial on how to carry out clinical studies in order to diagnose cerebrovascular disease. In fact each separate chapter draws on the specific interests of its own author, describing a range of techniques, some fairly routine but many semi-experimental, building on the experience and interests of a particular vascular laboratory. The disparate views of experts are particularly evident in places here, for example when setting out standardized methods for carotid evaluation. Dipping into these chapters the reader may find experimental studies in 3D flow patterns at the carotid bifurcation, a discussion of asymptomatic carotid stenosis, and a description of carotid artery pseudo-occlusion. Seven of the chapters review the developing area of transcranial Doppler, usefully relating MRI and angiographic images to the transcranial ultrasound studies, where these may be rather difficult to interpret. The high quality paper used enhances the grey-scale clarity both of the MR and x-ray images, and also of the numerous ultrasound images and histological sections with which the chapters are illustrated. The book concludes with a group of five chapters under the heading New and Future Developments. It is here that the text finally breaks with the repetition haunting the earlier pages, and each chapter in this section stands usefully on its own. Here the reader can find a concise review of contrast imaging, a description of ultrasonic thrombolysis, and an overview of emboli detection using multiple gated Doppler. Essential to the use of the volume is its extensive index, which has been prepared thoroughly and with care.

This is a book for the expert, not the novice. It assumes its reader to be active in cerebrovascular ultrasound, or in a laboratory devoted to vascular physiology. It expects the time and the mind to develop techniques and understanding using the expert experience in advanced laboratories around the world. It would be a worthy volume in an academic department library where innovation is part of the ethos. On the other hand I suspect that this book would be rarely taken down from the shelf of most vascular laboratories, where to be used and useful a book needs a more focused and coherent text than is contained here.