Lung Imaging Acquisition with Electrical Impedance Tomography: Tackling Common Pitfalls

The appropriate application of electrical impedance tomography and the interpretation of its data requires a basic understanding of electrical engineering and bioimpedance. In an electrical circuit, the flow of electricity depends on the voltage and the impedance. The flow of electrical current can be direct (steady flow) or alternating (backward and forward). In a condition of steady flow, the basic concept of an electric circuit is analogous to the movement of water in a pipe,²⁰  where the voltage represents the pressure exerted across the pipe (i.e., “pressure” on charged particles), the current is the water flow (i.e., the amount of charged particles per unit of time), the resistance is the difficulty imposed to water flow (i.e., opposition to the electrical current by a resistor).²⁰  In an alternating current system, in addition to the resistance to water flow, the effect of impedance can be represented by an elastic pipe, able to store and release energy depending on the variations in pressure and flow.

When biologic tissue or fluid is subjected to applied current, its electrical properties dictate the resulting voltage. These electrical properties are primarily resistance and capacitive reactance. The total opposition to an alternating current through tissues is termed impedance, an abstract physical variable, mathematically complex, that accounts for both resistance and capacitive reactance. The capacitive reactance opposition to an alternating current is inversely proportional to its frequency. Impedance can be expressed as a modulus or magnitude, denoted by |Z| and measured in ohms, and a phase, the difference in phase between current and voltage waveforms, denoted by φ and measured in degrees.^4,20 

Biologic tissues and fluids have different electrical properties due to their intracellular and extracellular composition, generating different voltages when current is applied. The extracellular and intracellular spaces have salt ions, and thus are efficient electricity carriers, whereas the cellular lipid membranes work as insulators.^4,20–22  “Impeditivity,” “resistivity,” and “reactivity” are the material properties adjusted for a unit of tissue (for example, impeditivity = impedance × area/length). However, several publications may address them as “impedance,” “resistance,” and “reactance.”

In the case of electrical impedance tomography, an alternating electrical current is applied via surface electrodes, then the subsequent voltage is measured. The current electrical impedance tomography devices estimate the variation of impeditivity, and the applied alternating current has low intensity (5 to 10 mA) and high frequency (50 to 125 KHz). It is harmless to biologic tissues and imperceptible to patients.^{1,2,4,5,13,20,22} 

The first electrical impedance tomography image reported was of a human forearm in the early 1980s by Barber et al.²³  Since then, electrical impedance tomography has quickly emerged as an imaging technique for other body tissues, including the lungs.²⁴  This review pertains to the application of electrical impedance tomography in lung imaging.

Over the years, electrical impedance tomography devices have become portable, commercially available, and less expensive.^5,15  In 2017, the U.S. Food and Drug Administration (Silver Spring, Maryland) approved electrical impedance tomography as a class II device that provides information on variations in regional air content across a section of the thorax in adult patients requiring mechanical ventilation. Electrical impedance tomography is not recommended in patients with pacemakers, defibrillators, and neurostimulators because it is unknown whether electrical impedance tomography interferes with their function (fig. 1). Moreover, the magnetic resonance imaging compatibility of electrical impedance tomography has not been tested.

Depending on the manufacturer, the chest perimeter, and the shape of the belts, the number of electrodes varies from 8 to 32.⁵  In general, an approximately 4-cm-wide electrodes belt is placed around the circumference of the chest wall for an adult patient. An important limitation of electrical impedance tomography is that the current moves diffusely within body tissue and tends to remain closer to the injecting electrodes (capturing information that is not of interest) instead of flowing deeper into the area of interest.^16,25  Different ways of current injection have been tested to monitor this phenomenon. The current can be injected simultaneously by a pair of electrodes or by multiple electrodes. The most common is a pair of electrodes injecting alternating electrical current around the chest wall, and, simultaneously, all electrodes measure the surface potentials. Depending on the device, the measurements from the electrodes injecting current may or may not be used to reduce the contact impedance influence on the image. Depending on the manufacturer, the pair of electrodes can be adjacent or have an adjustable skip, usually, three interposed passive electrodes.⁵ 

The thoracic plane on which the electrode belt is placed is crucial for electrical impedance tomography measurement and its reproducibility. Transverse plane placement is the most common option, and, before each belt placement, one should determine if the selected thoracic plane will cover the event of interest.⁵  The transverse thoracic planes can be divided craniocaudally, being the most cranial level right below the armpit at the third to fourth intercostal space, the middle level at the fifth intercostal space, and the most caudal level at the sixth and seventh intercostal space. If the electrode belt is placed cranially, the electrodes will be more distant, reaching obliquely dorsal portions of the lungs, which are regions at higher risk of atelectasis when compared with other regions of the lungs in patients with acute respiratory failure. The middle level (fifth intercostal space) is the thoracic plane suggested for placement by most authors, because it is considered the region where the intrathoracic structures are less affected by changes in position.^26–28  If belts are placed below the sixth intercostal space (caudal), the diaphragm is more likely to disrupt the impedance measurements (fig. 2).

Depending on the craniocaudal level, the relative amount of lung tissue is different; therefore, the electrical impedance tomography measures will also differ.²⁶  Krueger-Ziolek et al.²⁷  compared electrode placement at different thoracic planes (cranial, middle, and caudal) in healthy subjects during pulmonary function testing maneuvers. In their findings, variation in impedance amplitude was attributed to shifts in the lung tissue within the chest, either cranially during a maximum expiration to the residual volume, or caudally when the abdominal volume is shifted during maximum inspiration to total lung capacity. It is essential to acknowledge that multiple factors can impact what is being measured in a determined plane, not only the lung tissue shift, e.g., changes in ventilation, recruitment, or derecruitment, the shape of the chest, elevated diaphragm, the geometry of vascular and airway trees, and the blood volume within the chest.^27,29–31  If the belt is removed, marking the skin where the belt was placed for anatomical reference can help ensure the reliability of repeated measurements.

The order and the distance within the electrodes in the belts are preset in commercially available devices, and the belts come in different sizes to match the patient’s chest perimeter.⁵  The interelectrode spacing depends on the manufacturer, and the electrodes are usually equidistant. The electrode positioning should match the mathematical model of each electrical impedance tomography device designed to account for potential gaps in the sternal region. The gap will depend on the chest perimeter and the belt size. However, none of the models are planned to account for overlapping electrodes or rotated belts. Belt rotation may happen, especially in the neonatal population, and Sophocleous et al.³²  showed that a belt rotation higher than one electrode space significantly altered calculated electrical impedance tomography measurements.

Electrode-skin contact should be closely checked. The skin area selected for the electrode placement should be intact because the electrodes need direct contact with the chest wall to minimize the risk of poor-contact electrodes.^16,33  The application of electrodes over wounds, rashes, or burns is not recommended. Moreover, dressings should not cover the skin because they are nonconductive. The presence of hair can make electrode adhesion difficult, and ultrasound gel can be applied.¹⁴  Conductive suture wires, breast implants, and chest tubes can also interfere but rarely impede meaningful measurements.⁵  Electricity-conductive fluids (i.e., excessive sweat, saline, or other electrolytic solutions) should be avoided close to the electrodes. The electrodes are closer in neonates and preterm infants, so body lotions should be avoided in the chest area before assessments.^18,34 

The ability to distinguish between two close objects can define the electrical impedance tomography image resolution.³⁵  It depends on the amount of information available, which is proportional to the squared number of electrodes. More electrodes can provide a better spatial resolution and allow the exclusion of electrodes with poor contact (fig. 3). Once the device is on, the user should always check for loose electrodes because the image reconstruction assumes that electrode-skin contact impedance remains stable during measurements and the changes measured as due to internal variations in resistivity.³⁵ 

The sequence of current injections and voltage readings compose the frames that will generate the impedance image. Each complete acquisition cycle defines a frame (raw measurements).^19,28,36  Current electrical impedance tomography devices can generate up to 50 frames/second, resulting in high temporal resolution.^5,22 

Different from radiation-based images, applied current propagates in a diffuse pattern and expands to off-plane regions, and changes in these regions can affect the electrical impedance tomography measurements. In an electrode array at the same thoracic plane, the electrical current density of a specific region will determine how much that region influences the electrical impedance tomography image. Thus, in the craniocaudal axis, the thickness of the tissue slice with the highest electrical current density, centered in the plane of the electrode, is between 5 and 12 cm. If the electrodes are arranged in a zigzag pattern, this thickness changes and depends on the electrodes distancing from each other. Unlike the precise shape and thickness of cross-sectional slices from computed tomography and magnetic resonance imaging, the electrical impedance tomography image is a simplified reconstruction with a low spatial resolution.^{16,19,36–39}  The existing manufacturers prefer the two-dimensional images because they are easier than the three-dimensional images to reconstruct. Electrical impedance tomography image reconstruction is considered an inverse and ill-posed problem: inverse because it estimates the observed object information based on its measurement, and ill-posed because it does not have a unique or stable solution.¹⁶ 

In the screen of commercial electrical impedance tomography devices, the image is presented in the transverse plane, and the orientation of the image depends on the positioning of the electrodes and will stay the same independent of the posture, unless in the device where a screen command gives you the option to actively change the patient position between the supine/prone options.

The time-difference reconstruction is a strategy to minimize mathematical model intrinsic limitations. Differential electrical impedance tomography images only show regions with changes through time; unlike radiation-based imaging, “stable” findings such as consolidations, pleural effusions, or large bullae are not visualized. The differential image is based on a time-difference reconstruction between current and reference frames. A common reference measurement is the average of chosen frames (baseline). The color coding varies within manufacturers.^5,22  Multiple clinical tools are available depending on the electrical impedance tomography device. The tools can monitor the differential images through time and, for example, provide functional images of overdistension, collapse, compliance, and changes in the regional distribution of the impedance variation.

During a tidal volume breath, the electrical current has to move through thin and stretched interalveolar septa, resulting in an increase in resistivity during inspiration (maximum impedance), followed by a decrease in resistivity during exhalation (minimum impedance). The tidal impedance (ΔZ = maximum Z – minimum Z) measured at the body surface correlates with the tidal volume.^40–42  Moreover, observing the end-expiratory lung impedance at the bedside suggests shifts in functional residual capacity after changes in positive end-expiratory pressure (PEEP)^43,44  (fig. 4). Further comments on end-expiratory lung impedance analysis and limitations will be discussed in upcoming sessions.

The global impedance can be segmented into regional impedance waveforms that represent regions of interest. Commercial electrical impedance tomography devices have predetermined region of interest options during online monitoring, such as right and left, ventral and dorsal, quadrants, or up four layers (from nongravity to gravity- dependent regions).

Head-of-bed level influences the global and regional lung volumes and modifies the electrical impedance tomography waveforms (fig. 5).^15,45  As discussed before, multiple factors influence the measurements; thus, we advise the user to annotate and keep the head-of-bed constant while monitoring other events of interest, such as the response to different ventilatory modes, lung recruitment maneuvers, and optimal PEEP trials. Additionally, when monitoring ventilation with electrical impedance tomography in different body positions (supine, prone, right or left down decubiti, for example), the user should account for the physiologic effects of body position. Changes in body position may influence the regional distribution of ventilation, regional compliance of the respiratory system, and aeration⁴⁶  (fig. 6).

The diaphragm can interfere with the electrical impedance tomography measurement by causing a decrease in impedance variation (lower resistivity tissue) placed right next to a higher impedance variation of aerated lung tissue when compared with the diaphragm (higher resistivity tissue), misguiding the interpretation of impedance image.^26,28  Similarly, pendelluft⁴⁷  (movement of gas within lung regions, causing local deflations) and pleural effusion⁴⁸  can generate decreased impedance variations.

Extrathoracic events can affect the end-expiratory lung impedance waveforms and therefore mislead the estimation of lung aeration, e.g., abrupt patient or bed movements, pulsating air suspension mattresses,^18,34  and intravenous fluid bolus^49,50  (fig. 7). The administration of a saline solution bolus or diuretics can respectively decrease or increase the end-expiratory lung impedance without meaning changes in lung aeration. To avoid inaccurate measurements derived from end-expiratory lung impedance, users should refrain from considering it if there are significant ongoing changes in fluid balance.^34,50,51 

Electrical impedance tomography should be connected before the intervention of interest is performed. For example, electrical impedance tomography connection should occur before a lung recruitment maneuver to exclude a pneumothorax^52,53  and before stepwise PEEP titration.^54,55 

A stationary signal (e.g., stable tidal impedance returning to the same level of end-expiratory lung impedance) requires less recording time for offline analysis. If the patient spontaneously breathes, more time is needed to determine a reliable average signal (fig. 8). A minimum stabilization time of 5 to 10 min is recommended after an intervention of interest for the electrical impedance tomography to detect possible alterations.⁵⁶  Caruana et al.⁵⁷  suggested that at least 15 min is needed for ventilation distribution to stabilize when the body position is changed.

Prolonged monitoring periods are at a higher risk of electronic drifts and significant changes in the electrode contact impedance that cannot be fully compensated by the differential imaging method leading to lower quality reconstruction.³⁵ 

Table 1 summarizes the pitfalls, preventive approaches, and proposed solutions.

Electrical impedance tomography manufacturers incorporate automatic postprocessing tools for fundamental image analysis. These include algorithms to segment (separate) the functional lung maps from the heart and forms of regional partitioning (e.g., quadrant of the lung, ventral vs. dorsal) of the measured variables. In certain circumstances, offline analysis of saved raw data may be required. Several aspects of this process need to be addressed.

To recall and retrieve the actual acquisition, it is imperative to note all the events of interest (changes in ventilator settings or position, applied interventions), fluid administration and balance, and the occurrence of disconnections or artifacts.

For each individual subject, ventilation maps are usually filtered for noise by excluding pixels with values less than a certain percentage of the maximum (e.g., 5% or 10%). In patients with severe obesity, a thicker fat coating between the surface electrodes and the lungs will lead to a smaller image area. Redefining the regions of interest will help capture only those pixels carrying meaningful changes in impedance for the analysis.

Values of ventilation in each pixel are typically expressed as a percentage of the lung map total, thus representing a relative distribution for each subject. Correction is required to compare absolute values between subjects or interventions with externally obtained values of tidal volume (e.g., from the ventilator). For example, the end-expiratory lung impedance significantly correlates with end-expiratory lung volume measured by other methods (nitrogen washout, spirometry, helium dilution); however, the absolute volume calculations derived from end-expiratory lung impedance remain debatable.^31,44,58,59 

Frequent limitations of electrical impedance tomography’s broader application are the device setup and the understanding of the data provided.⁶⁰  Data sharing and interpretation by third-party specialist centers would boost the number of assessments, systematize the analysis, reduce misinterpretation, and promote education. Multicenter collaboration is crucial to guaranteeing electrical impedance tomography’s proper use, and machine learning can manage higher volumes of electrical impedance tomography data.⁶¹ 

The real-time monitoring of regional ventilation with electrical impedance tomography is invaluable for the acute care clinician. Research addressing the clinical impact of electrical impedance tomography applications is becoming more frequent.^62,63  However, a comprehensive understanding of the pitfalls of monitoring, data processing, and data interpretation in electrical impedance tomography is necessary to ensure accurate and reliable conclusions.

The authors thank Xiaojing Ma, M.D., Libin Ma, M.D., Changsheng Zhang, M.D., Ph.D., Caio C.A. Morais, R.T., P.T., Ph.D., Stefano Cenci, M.D., Dario Winterton, M.D., Hatus Wanderley, Kyle Medeiros, B.S., Carlo Valsecchi, M.D., Stefano Gianni, M.D., Raffaele Di Fenza, M.D., Bijan Safaee Fakhr, M.D., Susimeire Gomes, Ph.D., Adriana Hirota, R.T., P.T., Ph.D., and the Respiratory Care Service at Massachusetts General Hospital, Boston, Massachusetts.

Support was provided solely from institutional and/or departmental sources.

Dr. Alcala provides clinical support to customers for Timpel S.A. (electrical impedance tomography; São Paulo, Brazil). Dr. León Bueno de Camargo reports that his research group has received grants from the Timpel S.A. (electrical impedance tomography) in the past 5 yr. Dr. Britto Passos Amato reports that his research laboratory has received grants in the past 5 yr from the Covidien/Medtronics (mechanical ventilation; Dublin, Ireland), Orange Med/Nihon Koden (mechanical ventilation; Santa Ana, California), MagnaMed (São Paulo, Brazil) and Timpel S.A. (electrical impedance tomography). Dr. Amato is also a minority shareholder in Timpel S.A. The other authors declare no competing interests.