Introduction

For many years, clinicians have been looking for a proper way to evaluate the quality of ventilatory assistance for the individual patient. While some parameters, e.g., (dynamic) respiratory compliance, lower inflection point, pressure–time relationships, are available, there has been growing awareness that such global parameters may not be sufficient, and that in order to prevent lung damage, we will need to monitor regional lung mechanics at the bedside. A technology suitable for this task seems to be electrical impedance tomography (EIT) [1], which has recently received increasing attention from the intensive care community [25].

EIT is a noninvasive, radiation-free, and real-time imaging modality which is easy to use, not harmful to the patient, and offers regional information about ventilation and possibly eventually perfusion not available with any other monitoring modality. This method relies on the visualization and quantification of tissue impedance determined by injecting small electrical currents and measuring the resulting voltages at the surface of the torso. Owing to its capability to show changes in tissue impedance very quickly, it may be a valuable tool to adjust ventilator settings for the individual patient at the bedside [6, 7].

The literature on EIT is expanding rapidly [8] and new commercial devices are becoming available. Therefore, an increasing use in the clinical setting can be forecasted.

This review intends to provide a basis for understanding this development by summarizing the state-of-the-art and presenting a methodology to classify older findings and recent research activities.

Basic principle of EIT

In biological tissues, the conduction of electrical current is based on the local amount of fluid and ion concentration. Considering the lung, we note that the electrical current cannot flow through the alveoli, but only around them through the alveolar septa. While an increase in alveolar volume stretches and prolongs the average current pathways increasing the overall electrical resistance, expiration reduces the overall tissue resistance due to shorter and thicker current pathways.

In EIT, the distribution of local tissue resistance (or, more generally, impedance) within a cross section of the body is displayed as a tomographic image. As impedance Z is defined as voltage V over current I, regional impedance may be computed from measurements of currents and corresponding voltages on the surface (Fig. 1).

Fig. 1
figure 1

EIT measurement scenario with electrodes equidistantly attached to the torso of the patient, and with some front-end and some visualization electronics

At any time, there are an active pair of current-injecting electrodes and a separate pair of electrodes for voltage measurements. A typical setting is to inject a small alternating electrical current, e.g., 5 mA at a frequency of 50 kHz. To become independent of skin and electrode impedances, EIT expands the classical 4-electrode bioimpedance principle by continuously rotating the location of current injection and voltage measurements around the body on a plane, thus spanning a cross section. Typically, for lung monitoring the 4th–6th intercostal space is used.

In such a circumferential electrode arrangement, there are different ways to combine electrodes. The most popular configuration is referred to as “adjacent drive configuration” and combines neighboring electrodes forming adjacent pairs of electrodes [9]. Figure 2 illustrates the principle of rotating the locations of current injection (I) and voltage drop measurements (V) in this electrode configuration.

Fig. 2
figure 2

N = 16 electrodes in an adjacent drive configuration. Within a few milliseconds, both the current electrodes and the active voltage electrodes are repeatedly rotated around the body. (Cross section of torso adopted from [10])

In the left figure, an alternating current I is injected at time t between electrodes 1 and 2 while corresponding voltages are sequentially measured at all remaining electrode pairs (e.g., N − 3 = 13 combinations). Afterwards, at time t + ∆T the location of current injection is rotated to the next pair of electrodes (electrodes 2 and 3, right figure) and again the corresponding voltages are sequentially measured at the remaining 13 electrode pairs. One image can be computed as soon as all electrode pairs have been used to inject current.

Temporal resolution may be relatively high in EIT (13–50 frames/s), with the potential to cover both ventilation and perfusion in real time. In contrast, spatial resolution of the images is significantly lower and at best limited to the distance between electrodes (ca. 2–3 cm).

Absolute and relative images

Obtaining accurate absolute EIT images is difficult because this requires detailed knowledge about geometry and absolute electrode positioning [11]. To circumvent this problem and improve the quality of EIT images, Barber [12] and later Hahn et al. [13] introduced the concept of “relative” EIT imaging. Here, the average impedance over a certain time window is subtracted from the actual impedance value which establishes a contrast amplification and cancels most of the geometric artifacts common in absolute EIT images. However, the price paid is the loss of absolute impedance information, i.e., only impedance changes ∆Z n remain visible. Whereas the information on regional ventilation is preserved, a differential diagnosis of (non-ventilated) constant-impedance areas is not directly possible anymore (like distinguishing between overdistension, pneumothorax, excess in pleural fluid, or atelectasis).

A common way to visualize EIT images is to use a rainbow-color scheme [14], where red represents inspiration (highest relative impedance values), green codes for the breathing baseline, and blue (lowest relative impedance values) expiration. Figure 3 presents an example of the resulting colored image stream.

Fig. 3
figure 3

Sequence of a rainbow-colored resting breathing cycle using relative EIT images based on ΔZ n , according to [13]

In a recent consensus paper [15], a different color scale was introduced where red implies low impedance (liquid) and azure stands for high impedance (air) with medium levels in black. Some of the commercial devices which became recently available use similar color schemes spanning from black to ultramarine blue-white for ventilation.

Functional imaging (fEIT)

The concept of functional EIT imaging was originally introduced in 1995 [13, 16]. Basically, this concept aims at imaging global and local ventilation activity over a certain period of time which can be selected by the user (e.g., 30 s). The authors proposed to quantify amplitudes of ventilation activity by computing local standard deviations at each pixel and plotting them at the corresponding image position. Such fEIT images (Fig. 4) provide a steady-state map of activity (mainly regional ventilation) causing impedance changes in the corresponding cross section of the body.

Fig. 4
figure 4

Differences in geometry and corresponding rainbow-colored fEIT images of a human (left) and a pig (right). Strong impedance in red indicates regions of ventilatory activity, whereas low impedance changes (blue) point at no regional ventilation. Note the different meaning of the color scheme as compared to Fig. 3. Modified from [17]

fEIT images are especially useful to position cursors and define regions of interest (ROIs), like specific pixels, layers, or quadrants. Furthermore, they allow one to study the ventilation distribution (e.g., regarding its homogeneity). Hence, positioning cursors at selected ROIs in fEIT images in combination with looking at the corresponding time courses of local impedance curves has facilitated the specific study of the ventilation distribution (Fig. 5). If selected properly, local impedance changes may not only display pulmonary but also cardiac-related impedance changes.

Fig. 5
figure 5

Example of global and local impedance time courses (right curves) in a human subject. While the image stream of relative EIT frames is displayed as a movie (upper left corner), the fEIT image may conveniently be used for positioning cursors on selected ROIs. Note that in this example, ∆Z local,3 shows cardiac activity superposed on the ventilatory signal, whereas ∆Z local,4 positioned in the lower right corner shows no ventilatory activity at all

Techniques for ventilation monitoring

Relation between global and regional ventilation

As mentioned above, fEIT images are thought to capture and image local ventilation activity. Hence, summing up all pixels displayed in an fEIT image is referred to as global ventilation activity (GVA). A similar index named global tidal variation may be computed by summing all local ventilation-induced impedance changes ∆Z local,max − ∆Z local,min.

The correlation of global ventilation with global impedance measurements ∆Z global resulting from EIT image streams has been investigated by many researchers, e.g., using a large syringe for a predefined gas or fluid volume injection (air, albumin solution) into the lungs [18], and using plethysmography [19].

An interesting question is whether changes in end-expiratory lung impedance (EELI) correlate with changes in end-expiratory lung volume (EELV). In dogs, a linear increase in impedance was demonstrated when the lung was inflated with a calibrated syringe [18]. Similarly, using multi-breath nitrogen washout as a reference, a good correlation (R 2 = 0.95) between ∆EELI and ∆EELV was found in a small human positive end-expiratory pressure (PEEP) trial involving 10 patients [20]. Using a super syringe as a reference, this question was also investigated in a lavage model involving 14 pigs [21]. In this trial, the authors quantified changes in regional functional residual capacity (FRC) based on local ΔEELI in four different layers and reported a good correlation between (global) EELV and EELI (R 2 > 0.95). In a clinical study including 12 patients, a correlation of R 2 = 0.92 between changes in EELI and spirometric measurements of tidal volume was demonstrated [22]. For illustration, Fig. 6 shows the time course of global impedance ∆Z global during a recruitment maneuver.

Fig. 6
figure 6

Time course of global impedance ∆Z global showing the effect of a 40-s recruitment maneuver

However, the results presented by Bikker et al. [23] from a human trial including 25 ARDS patients only showed a moderate correlation and therefore do not support the assumption of a purely linear relationship between global ∆EELI and ∆EELV. This weaker correlation may be based on severe changes in lung geometry due to heterogeneously impaired lung regions (atelectasis, overdistension, normally aerated regions). On the other hand, the conclusion of this study may be somewhat misleading as the chosen belt position (7th intercostal space) might be low enough for the diaphragm to enter the measurement field for low PEEPs. This could explain the low correlation. Thus, the overwhelming scientific information today is that ∆EELI can be scaled and converted to volume (ml) by using the global tidal variation, accepting that changes in intrathoracic blood volume occur when changing PEEP and this must be corrected for by taking into account the global tidal variation at the different PEEP levels.

In 2002, the correlation between regional ventilation and electron-beam computed tomography (EBCT) was studied in six pigs [24]. A good correlation (R = 0.81–0.93 depending on the lung region) between EIT and EBCT measurements was reported. Shortly after, similar findings were presented in a human study of 10 patients during slow inflation maneuvers [25], with a correlation of R 2 = 0.96 between regional air content changes determined by CT and impedance changes measured by EIT. Other studies validating regional EIT information include dynamic X-ray computed tomography (CT) in pigs [26], single-photon emission computed tomography (SPECT) [27], and positron emission tomography (PET) [28]. In conclusion, at present most researchers no longer question the fact that EIT correlates well with regional ventilation in the torso.

∆fEIT imaging

The concept of ∆fEIT imaging was introduced in [29] and [30]. Basically, an fEIT image representing a specific situation at time t 2 is compared to an fEIT image at a previous time t 1 and the changes are contrast-amplified by a pixel-by-pixel subtraction. In order to obtain an approximation for changes in regional tidal volume, the difference is normalized by tidal volume V T divided by GVA (Fig. 7).

Fig. 7
figure 7

Basic concept of ∆fEIT imaging, modified from [29]. In this example, PEEP has been reduced from 15 to 10 cm H2O. While global V T increased from 320 to 329 ml, dorsal atelectasis developed and the distribution of ventilation became much more inhomogeneous indicating that this PEEP was already too low. Note that the two color codes have different meanings

It is anticipated that this specific technique provides clinically important information on regional gain or loss of ventilation when changing ventilator settings. For example, to find an optimal PEEP in a decremental PEEP trial, global information like the tidal volume may be misleading. As shown in Fig. 7, global tidal variation still increased while reducing PEEP from 15 to 10 cm H2O already caused a significant loss in dorsal ventilation, but was superimposed by a larger gain in ventral ventilation. Although the following statement remains debatable [31], it is widely accepted that a very inhomogeneous ventilation would not be beneficial for severely sick patients, but will increase the risk of ventilator-induced lung injury (VILI).

Recently, ∆fEIT imaging was successfully applied within a small human patient trial [32].

Derived indices

Although EIT images and movies match well with the visual system of human operators, it is rather difficult to quantify information based on EIT image data. Since quantification is a prerequisite for therapeutic decisions, the following section reviews the literature regarding established numerical indices for EIT images.

Impedance ratio

In the year 2000, a new index called the impedance ratio (IR) was introduced which divided the ventilation activity in the upper (ventral) region VAventral of fEIT images by the dorsal ventilation activity VAdorsal [33]. This index turned out to be a sensitive parameter for following recruitment and derecruitment in animal trials.

Center of ventilation

The idea of using the center of gravity (CoG) as an index to characterize the shape of functional EIT images was originally introduced in [34]. Considering the application, this index was named center of ventilation (CoV). This concept can be extended to characterize shapes of fEIT images as well. In 2007, it was demonstrated that y cog may serve as a sensitive index to describe vertical shifts in ventilation due to opening and closing of the lung during incremental and decremental PEEP trials [28].

Regional delay index

The concept of imaging regional delays in ventilation during slow inflation maneuvers was introduced in [26]. Basically, the so-called regional ventilation delay (RVD) index quantifies the delay time needed for the regional impedance–time curve to reach a certain threshold of the maximal local impedance. Hence, it quantifies not just regional, but also temporal information from the EIT images, mainly on cyclic opening and closing. Recently, this concept was extended by providing maps of RVD standard deviations which correlate well with lung inhomogeneity [4, 35].

Homogeneity index

Another simple index called the global inhomogeneity index (GI) was proposed in [36, 37]. To calculate the GI, one has to compute the median value of regional impedance changes from the tidal image, then calculate the sum of differences between the median and every pixel value, and finally normalize by the sum of impedance values within the lung area. As reported, this index was able to differentiate ventilation inhomogeneity during one-lung ventilation. The performance of this index was investigated on a total of 50 patients [36], out of which 40 patients were tracheally intubated with double-lumen tubes (test group) and 10 patients serving as a control group were under anesthesia without pulmonary disease. One-lung ventilation was clearly distinguishable from double-lung ventilation for all patients. Furthermore, the GI index was found to be usable for PEEP titration in a pig trial where it correlated well with maximum compliance [37].

Quantification of regional overdistension and atelectasis

Another index which allows one to quantify regional overdistension and atelectasis was introduced in [38] and recently extended [39]. The methods assumes a uniform pressure distribution within the lung and introduces the concept of pixel compliance, i.e., ∆Z/(P plateau − PEEP). By comparing the pixel compliance to the maximum regional compliance, an index for overdistension or atelectasis can be derived.

Application examples

Quantifying the effects of suctioning and disconnection

In an animal model and in a small clinical trial (13 ICU patients), it was recently demonstrated using EIT that functional residual capacity (FRC) is affected during bronchoscopic suctioning [40]. Also, the effects of disconnection and endotracheal suctioning in nine ventilated pigs were described [41]. Here, EIT was used to quantify the changes in regional compliance affecting mainly the dorsal lung regions.

In 2007, Wolf et al. [42] reported on using EIT to quantify the effect of derecruitment and suctioning in six children with acute lung injury (ALI). Similarly, a substantial loss of lung volume during endotracheal suctioning and the dynamics of recovery were demonstrated by Tingay et al. [43] in six 2-week-old piglets and by van Veenendaal et al. [44] in 12 preterm infants.

Unilateral intubation and placement of endotracheal tube

As reported in [45], EIT may be useful to position a tracheal tube. In a study of 40 patients, double-lumen tubes were employed to establish selective one-lung ventilation for thoracic surgery. Fibre-optic bronchoscopy was used to confirm double-lumen tube position. It was concluded that EIT allows the accurate online display of left and right lung ventilation and enables the clinician to check the proper positioning of the tube.

Regional pressure–volume curves

It is well known that surfactant-depleted lungs develop a sigmoidal pressure–volume (PV) curve characterized by a dedicated lower inflection point (LIP) and an upper inflection point (UIP) [46, 47]. Regional PV curves using EIT were investigated in [48] and results from four horizontal layers of diseased animal lungs were presented. Here, the underlying assumption was that ventilation pressure is effective and constant throughout the lung. Furthermore, as introduced above, ∆Z local was assumed to represent regional lung volume changes [24, 25]. The resulting opening pressures varied significantly in the different lung layers, with the highest pressures found in the dorsal lung layer. This work was extended by proposing to image the regional distribution of LIP and UIP in different parts of the lung [49, 50]. The regional differences of LIP and the homogenizing effect of high-frequency oscillatory ventilation (HFOV) on the distribution of regional PV curves in ALI pigs were reported in [51]. In [52], the use of regional PV loops and compliance for titrating PEEP was proposed and demonstrated in 16 neonatal piglets. Similarly, on the basis of the results of 16 ALI/ARDS patients, it was recently demonstrated that proper analysis of regional PV loops and compliance may be useful to identify individual changes in gas distribution, lung recruitability, and help to individually select PEEP [53].

Pneumothorax monitoring

Although uncommon, a pneumothorax is a potentially dangerous incident that may arise unexpectedly. By definition, in such an incident air is released into the thoracic cavity. As demonstrated in 2008, EIT can be used to quickly detect the development of an artificial pneumothorax in a controlled experimental setting [54]. Shortly thereafter, the time course of an accidental pneumothorax which unexpectedly occurred during an open lung animal experiment was reported [55].

However, while detection of accidental pneumothorax certainly is an advantage, long-term pneumothorax monitoring using EIT may be impractical as there may be a chance of developing an ulcer and necrosis in dorsal regions due to electrode belt pressure.

Pulmonary edema

Few studies have investigated whether EIT can be used to quantify the amount of fluid volume in the lung. In 1999, it was investigated how the IR index compares to extravascular lung water (EVLW) determined by the thermal dye double indicator dilution technique (TDD) [56]. Fourteen ALI/ARDS patients were included into the study and a significant correlation between changes in EVLW as measured by TDD and EIT (R = 0.85; p < 0.005) was presented. Using an eight-electrode EIT device with reduced spatial resolution, a strong correlation between thoracic resistivity and removed pleural fluid in 11 pleural effusion (PE) patients was recently reported [57].

Recruitment and derecruitment

It is well known that gravity may cause dorsal atelectasis in diseased patients, while ventral regions may be normally ventilated and sometimes even overdistended. Hence, monitoring the transverse distribution of ventilation is an important application of EIT, especially as a function of PEEP.

For illustration, Fig. 8 demonstrates the shift of regional ventilation due to PEEP. In this animal, ALI was induced by repetitive lung lavage [58] with saline warmed to 37 °C.

Fig. 8
figure 8

CT scan and corresponding fEIT image with four layers in an ALI pig ventilated with FiO2 50 %, I:E 1:1 and ∆p = 10 mbar. Upper images (PEEP = 0 mbar): CT scan shows dorsal atelectasis. In the fEIT image, the main ventilation activity (54.8 % of all ventilation activity) occurs in layer 2. Lower images (PEEP = 20 mbar): main ventilation activity has shifted to layer 3 (47.3 %) of the fEIT image and ventilation is much more homogenous. At the same time, the corresponding CT scan shows no atelectasis anymore (data from the authors, not published)

Although some authors used the IR index to quantify shifts in ventilation [33, 59], it was shown in [29] that the center of gravity y cog may be employed to quantify such shifts during incremental and decremental PEEP trials. Similarly, by counting active pixels, shifts of ventilation activity due to PEEP changes were quantified in eight patients [60]. Furthermore, it was recently proposed to use EIT for local compliance estimation leading to means to access lung recuitability [61]. In [62], it was pointed out that y cog correlates well with oxygenating efficiency during high-frequency oscillatory ventilation.

On the basis of CT measurements in two patients, Costa et al. [38] introduced the concept of pixel compliance and presented an EIT-based technique to quantify alveolar collapse and regional overdistension based on these indices. Recently, Costa’s technique was successfully applied during a stepwise recruitment maneuver in nine ALI/ARDS patients [39].

Perfusion monitoring

As the perfusion-related changes in thoracic impedance are about one order of magnitude smaller than the changes induced by ventilation, it is much more difficult to extract information on stroke volume (SV), cardiac output (CO or Q), or lung perfusion.

Examples of early work on this subject were published in 1996 [63]. These studies focussed on the determination of stroke volume in healthy volunteers and used magnetic resonance imaging (MRI) as a reference. As has been demonstrated for idiopathic pulmonary arterial hypertension (IPAH) [64, 65], there may be more important connections between ventilation and cardiac-related impedance signals, as these patients expressed significantly lower impedance changes in the cardiac frequency band as compared to healthy controls.

At present, all attempts to quantify perfusion are still in the research stage and not yet available at the bedside. Some algorithmic questions remain to be answered, for which some approaches will be reviewed in the following subsections. However, the potential is large, as EIT-based perfusion monitoring would allow one not only to image local perfusion but also to establish local ventilation/perfusion (V/Q) mappings.

Separation of cardiac- and ventilation-related EIT signals

This approach relies on the separation of the ventilation-related signal (VRS) and the cardiac-related signals (CRS) based on frequency content and was originally introduced in the early 1990s [66, 67]. It has been successfully employed in a bilateral ventilation study to quantify left-to-right lung perfusion ratios in 10 patients scheduled for elective chest surgery [68]. By employing such frequency-domain filtering techniques on EIT measurements, Carlisle et al. recently reported [69] gravity-dependent differences in the distribution of perfusion and ventilation were in 26 preterm infants.

ECG gating

This technique amplifies the CRS by assuming that it has a deterministic characteristic and that VRS is completely uncorrelated with CRS [7073]. The QRS complex of the synchronously measured ECG signal is used as a trigger. By summing up the images over, e.g., 200 heart cycles starting at each QRS complex, the signal strength of CRS is selectively amplified by a factor of √200 ≈ 14 in comparison to the relative strength of VRS. Additional electrodes for ECG measurements may not be necessary, as the EIT electrodes themselves could be used. Note, however, that considering 200 heart cycles implies that a later visualization of CRS on screen would be delayed by 200/2 = 100 heart beats, which excludes real-time analysis.

Apnea methods

In this approach, ventilation is briefly interrupted to quantify impedance changes introduced solely by heart activity (CRS). On the basis of this method, it was recently investigated whether EIT is capable of monitoring stroke volume and perfusion [74]. In this study, cardiac preload, and thus pulmonary perfusion, was reduced by inflating a balloon of a Fogarty catheter positioned in the inferior caval vein or by increasing the PEEP. Stroke volume (SV) was derived using a pulmonary artery catheter and thermodilution method. A reasonable beat-to-beat correlation between stroke volume and the EIT global impedance signal during apnea was found. Shortly thereafter, it was shown that measurements of global impedance during ventilation and during apnea could be used to produce regional Z V/Z Q values in distinct ROIs [75]. Despite remaining limitations, it was pointed out that such EIT measurements may have the potential to eventually produce local V/Q mappings. However, although inducing a short apnea for such measurements may not severely harm the patient, it changes the ventilatory situation. Hence, it is not guaranteed that the same perfusion will occur during ventilation and during apnea.

Use of contrast agents

As was demonstrated in [24], hypertonic saline solutions may be used as an electrical impedance contrast agent. In this work, a Swan–Ganz catheter was positioned in a pulmonary artery branch and 5.85 % saline solution was administered as a bolus. EBCT scans and a radiographic contrast material were used as a reference. The authors were able to track the high salt concentration bolus in three pigs on its way from the right heart via the lungs to the left heart. Similar findings using a 10 % hypertonic saline solution and SPECT measurements for reference in four pigs were reported in [76]. A similar scenario was recently described in [77]. Although these results are preliminary, this approach seems to have some potential for quantification and future clinical application, as most patients on the ICU have a central venous catheter in place. However, a principal limitation is that the analysis is not beat-to-beat and delayed by the time-of-flight necessary for the passage of the contrast agent (about 10 s).

Separation based on principal component analysis

In 2008, a new algorithm based on principal component analysis (PCA) was introduced [78]. For convenience, Fig. 9 provides an example of successful separation of CRS and VRS using PCA in a pig trial. Note the heart-related activity in the right image which is not visible in the ventilation image.

Fig. 9
figure 9

Successful separation of an EIT image stream into a pulmonary-related EIT signal (left, rainbow color scheme) and an amplified cardiac-related EIT signal (right, black-red-white coloring) using PCA. The corresponding video is provided as electronic supplementary material 2

While being the first specific algorithm of its kind to separate the two signal components in a beat-to-beat mode, some of the underlying theoretical basis has already been presented in 2001 [79]. Provided there is enough computational power, this promising approach has the potential to be fast enough for beat-to-beat monitoring. For illustration, we provide another supplementary material 1 which is an example of successful CRS (right) and VRS (middle) separation from the raw image stream (left). However, at present the results are still preliminary and require confirmation in larger trials.

Outlook

For the transition of EIT from the research lab to the ICU, clinicians will require not only useful applications but also sufficient ease of handling. Important aspects of the user interface, such as electrode belts and ways to visually represent information on screen, will be key issues for this transition.

Especially with respect to the ALI/ARDS patient population, there is a strong need for consensus on hardware, color schemes, and imaging algorithms, like the one provided in [15], in order to be able to move on to multi-center trials while staying compatible.

Although regional ventilation will become the first clinical EIT target application, it is foreseeable that regional perfusion and regional V/Q mappings are next in line. However, other possible extensions of current planar monofrequent EIT include techniques like multifrequency tomography [80, 81] or three-dimensional impedance tomography [5, 82, 83]. Finally, while being a difficult algorithmic problem, absolute EIT (aEIT) imaging [84, 85] has the potential to differentiate between non-ventilated areas containing mainly pulmonary air (pneumothorax) or fluid (atelectasis, edema, pleural effusion). Eventually, aEIT may even allow one to quantify changes in FRC.

Conclusions

Electrical impedance tomography is a new monitoring modality that will change the way we treat patients by mechanical ventilation. It will allow us to tune and optimize regional ventilation of the individual patient which, until now, is difficult at the bedside. Furthermore, there is potential for future extensions, including regional perfusion monitoring, regional V/Q mapping and, possibly, quantifying pneumothorax, atelectasis, pulmonary edema, and possibly the onset of pneumonia at an early stage.

In 2004, Arnold anticipated that EIT would lead to the “holy grail of ventilation monitoring” [1]. On the basis of the evidence provided in this review, we believe that he was right, but possibly regional ventilation monitoring is just the dawn of a new era of opportunities.