Secondary Logo

Journal Logo

Estimation of Stroke Volume and Stroke Volume Changes by Electrical Impedance Tomography

da Silva Ramos, Fernando José MD*; Hovnanian, André MD, PhD*,†; Souza, Rogério MD, PhD; Azevedo, Luciano C. P. MD, PhD*; Amato, Marcelo B. P. MD, PhD; Costa, Eduardo L. V. MD, PhD*,†

doi: 10.1213/ANE.0000000000002271
Technology, Computing, and Simulation: Original Laboratory Research Report
Free
SDC

BACKGROUND: Electrical impedance tomography (EIT) is a noninvasive imaging method that identifies changes in air and blood volume based on thoracic impedance changes. Recently, there has been growing interest in EIT to measure stroke volume (SV). The objectives of this study are as follows: (1) to evaluate the ability of systolic impedance variations (ΔZsys) to track changes in SV in relation to a baseline condition; (2) to assess the relationship of ΔZsys and SV in experimental subjects; and (3) to identify the influence of body dimensions on the relationship between ΔZsys and SV.

METHODS: Twelve Agroceres pigs were instrumented with transpulmonary thermodilution catheter and EIT and were mechanically ventilated in a random order using different settings of tidal volume (VT) and positive end-expiratory pressure (PEEP): VT 10 mL·kg−1 and PEEP 10 cm H2O, VT 10 mL·kg−1 and PEEP 5 cm H2O, VT 6 mL·kg−1 and PEEP 10 cm H2O, and VT 6 mL·kg−1 and PEEP 5 cm H2O. After baseline data collection, subjects were submitted to hemorrhagic shock and successive fluid challenges.

RESULTS: A total of 204 paired measurements of SV and ΔZsys were obtained. The 4-quadrant plot showed acceptable trending ability with a concordance rate of 91.2%. Changes in ΔZsys after fluid challenges presented an area under the curve of 0.83 (95% confidence interval, 0.74–0.92) to evaluate SV changes. Conversely, the linear association between ΔZsys and SV was poor, with R2 from linear mixed model of 0.35. Adding information on body dimensions improved the linear association between ΔZsys and SV up to R2 from linear mixed model of 0.85.

CONCLUSIONS: EIT showed good trending ability and is a promising hemodynamic monitoring tool. Measurements of absolute SV require that body dimensions be taken into account.

Published ahead of print July 24, 2017.

From the *Department of Intensive Care and Anesthesiology Research Laboratory, Research and Education Institute, Hospital Sírio-Libanês, São Paulo, Brazil

Respiratory Intensive Care Unit, University of São Paulo School of Medicine, São Paulo, Brazil.

Published ahead of print July 24, 2017.

Accepted for publication May 9, 2017.

Funding: The experimental protocol was supported by Fundação de Amparo a Pesquisa do Estado de São Paulo (FAPESP) and by the Research and Education Institute, Hospital Sírio-Libanês.

Conflicts of Interest: See Disclosures at the end of the article.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website.

Reprints will not be available from the authors.

Address correspondence to Fernando José da Silva Ramos, MD, Research and Education Institute, Hospital Sírio-Libanês, Rua Cel, Nicolau dos Santos, 69, São Paulo 01308, Brazil. Address e-mail to ramosfjs@gmail.com.

An important goal of the management of critically ill patients is the optimization of hemodynamics. To this end, hemodynamic monitoring plays a central role. Hemodynamic monitoring can be used in patients with established organ failures due to shock or even before the development of organ dysfunctions, such as in high-risk surgical patients.1,2 In the last years, minimally invasive devices for hemodynamic monitoring have substituted the classic pulmonary artery catheter, because the use of pulmonary artery catheters led to higher morbidity associated with its invasiveness.3–7

Another major paradigm shift has been the appreciation that volume management can be guided by the fluid responsiveness concept, in which a variety of techniques are used to determine whether the administration of intravenous fluids will improve cardiac output (CO) or stroke volume (SV).8–10 Based on this concept, relative changes in CO are used to define fluid responsiveness and are common targets in clinical practice to optimize hemodynamics and avoid unnecessary fluid load. This strategy is supported by several randomized trials showing improvements in outcomes.11,12

Electrical impedance tomography (EIT) is a noninvasive imaging method that has been described to assess ventilation inhomogeneity, to evaluate recruitment and pulmonary collapse,13 and to detect pneumothoraces after placement of central venous lines.14 Recently, there has been growing interest in EIT to evaluate pulmonary perfusion and to measure SV.15–17

We hypothesized that relative changes in systolic impedance variation (ΔZsys) have the ability to track changes in SV efficiently in relation to a baseline condition. In addition, we postulated that the correlation of ΔZsys with absolute measurements of SV is affected by body dimensions, which are known to influence thoracic wall composition.18 Therefore, our aims were to (1) evaluate whether relative changes in ΔZsys were able to reliably track changes in SV; (2) investigate the relationship between ΔZsys and SV; and (3) assess the influence of body dimensions in the correlation of ΔZsys to SV.

Back to Top | Article Outline

METHODS

This is an experimental study with crossover design, in which adult female pigs were submitted to hemorrhagic shock and sequential fluid challenges to compare SV with ΔZsys and to evaluate fluid responsiveness.

The experimental protocol was performed at the Research and Education Institute of Sírio-Libanês Hospital (São Paulo, Brazil) and was approved by the Institutional Animal Research Ethics Committee (CEUA 2011/14). It was performed according to the National Institutes of Health guidelines for the use of experimental animals.

Back to Top | Article Outline

Instrumentation and Stabilization Period

Twelve adult female Agroceres© pigs (RG farm, Suzano, Brazil) weighing 36.8 ± 8.9 kg were fasted overnight with free access to water and premedicated with an intramuscular injection of midazolam (0.3 mg·kg−1) and acepromazine (0.5 mg·kg−1). Anesthesia was induced with thionembutal (12 mg·kg−1). Anesthesia was maintained during the study with midazolam (0.3 mg·kg−1·h−1) and fentanyl citrate (25 μg·kg−1·h−1), and muscular relaxation with pancuronium bromide (0.2 mg·kg−1·h−1). The animals were submitted to mechanical ventilation (Evita XL; Dräger, Luebeck, Germany) with the following baseline ventilatory settings: tidal volume (VT) 8 mL·kg−1, positive end-expiratory pressure (PEEP) 10 cm H2O, inspiratory flow of 40 L·min−1, respiratory rate of 24 inspirations per minute and later adjusted to maintain an end-tidal-carbon dioxide between 35 and 45 mm Hg, and inspiratory fraction of oxygen adjusted to maintain arterial saturation >93%.

A 5-Fr catheter (PV2015L20, Pulsiocath, Pulsion Medical System, Munich, Germany) was inserted into the left femoral artery and connected to the Infinity PiCCO SmartPOD (Dräger Medical, Luebeck, Germany) to monitor arterial pressure and pulse contour analysis. The left internal jugular vein was cannulated with a 7-Fr triple lumen catheter and connected to the Infinity PiCCO SmartPOD (Dräger Medical) to perform transpulmonary thermodilution CO (COTPT) measures and to perform drug infusions.

During the stabilization phase, animals received continuous infusion of lactated Ringer’s solution 5 mL·kg−1·h−1. After a stabilization period of 30 minutes, the protocol was initiated.

Back to Top | Article Outline

Hemodynamic Measurements

The Infinity PiCCO SmartPOD (Dräger Medical) system was used to perform COTPT measurements. SV by transpulmonary thermodilution technique (SVTPT) was also calculated. We performed COTPT measurements in triplicate using 15 mL of cold saline solution after 2 minutes of data acquisition in each phase for each ventilatory strategy.

Back to Top | Article Outline

Study Protocol

Data were collected in at least 4 different conditions: baseline, hemorrhage, and sequential fluid challenges with 500 mL of lactated Ringer’s solution until the animal became nonresponsive. The nonresponsive state was defined a priori and according to the current literature as an increase in SV of less than 10% of the previous SV measured.10,19 We performed a sensitivity analysis with the cutoff of 15%. In each one of these conditions, the animals were ventilated with 4 different ventilatory settings in a random order using different settings of VT and PEEP: VT 10 mL·kg−1 and PEEP 10 cm H2O, VT 10 mL·kg−1 and PEEP 5 cm H2O, VT 6 mL·kg−1 and PEEP 10 cm H2O, and VT 6 mL·kg−1 and PEEP 5 cm H2O. We anticipated that PEEP changes could cause more sustained hemodynamic changes. For this reason, we randomized the order of the ventilatory settings to avoid a possible systematic bias due to carryover effect. After 2 minutes at each ventilatory setting, hemodynamic and respiratory data were recorded. After a baseline data collection, hemorrhage was performed by withdrawal of 50% of estimated blood volume from each animal through the femoral artery line. Estimated blood volume was considered as 70 mL·kg−1 of body weight and hemorrhage interval varied from 20 to 30 minutes. After data collection during hemorrhage, a fluid challenge (named 1) of 500 mL of lactated Ringer’s solution was infused over 5 minutes, and, after 2 minutes of stabilization, data collection was performed. Other fluid challenges (named 2, 3, and 4) with the same characteristics were performed sequentially until the animal became nonresponsive. After data collection, the animals were euthanized while under anesthesia with potassium chloride overdose.

Back to Top | Article Outline

Electrical Impedance Tomography Data Acquisition

EIT is a noninvasive, radiation-free imaging technique used to estimate cross-sectional images of the intrathoracic regional impedances. Regional impedances depend mostly on tissue composition and their electrical resistivity.20 For example, the lungs have values of resistivity that range with inspiration from 7.2 to 23.6 Ωm (the higher the air content, the higher the local impedance). Blood conducts electricity more easily than the lungs (resistivity of approximately 1.6 Ωm) such that, following the cardiac systole, the excess of blood that enters the lung promotes a drop in impedance in proportion to the amount of blood received by the lungs. Therefore, the raw EIT signal combines information from both air and blood volume inside the lungs.21 Evaluation of pulmonary pulsatility by EIT is based on the principle that right ventricle SV during cardiac systole causes increases in both blood volume and intravascular pulmonary pressures, which leads to elastic distension in microvascular pulmonary bed and consequently local impedance reduction, that is known as systolic impedance changes. Efficient separation of these signals, for example, filtering out ventilation with electrocardiography gating (EKG gating), allows the estimation of ΔZsys, a measurement correlated with the right ventricle SV.

The EIT data were acquired at a rate of 50 images/s using the EIT Enlight (Timpel, São Paulo, Brazil). EIT and hemodynamic data were acquired simultaneously. A self-adhesive belt composed of 32 electrodes was attached in a transversal plane around the circumference of the thorax just below the level of the axilla. Small electrical alternating currents (5–12 mA; 125 KHz) were injected in a rotating sequence through pairs of electrodes, while noninjecting electrode pairs were used to measure differential voltages. Each complete acquisition cycle of 32 different pairs of injecting electrodes was processed into a relative EIT image representing the distribution of electrical impedances within the thorax (Supplemental Digital Content 1, Figure, http://links.lww.com/AA/B847). A total of 50 such EIT images were generated each second (frame rate = 50 Hz), representing the intrathoracic distribution of electrical impedances. Data were analyzed by dedicated software written in Labview (National Instruments, Austin, TX). The EKG gating enabled automatic segmentation of lung and heart regions of interest (ROIs) through the observation of the impedance curve orientation after the EKG R-wave (Supplemental Digital Content 2, Figure, http://links.lww.com/AA/B848): a decrease in ΔZsys indicates lung pixels, whereas an increase indicates heart pixels. The sum of all pixels values contained in the lung ROI was used as the measure of ΔZsys. All ΔZsys measurements were obtained immediately before thermodilution measurements.

Back to Top | Article Outline

Statistical Analysis

Variables are expressed as mean and standard deviation or median and interquartile rage when appropriate. The trending ability of changes in ΔZsys was assessed with 4-quadrant plot.22 We used repeated measures analysis with a linear mixed-effect model to estimate an marginal R2 from linear mixed model (R2LMM; variance explained by fixed factors) for the 4-quadrant plot.23 The receiver operating characteristic (ROC) curve was built to test the accuracy of ΔZsys to evaluate fluid responsiveness, defined as an increase in SV > 10% after a fluid challenge. We computed the area under the ROC curve using mixed-effect logistic regression (generalized linear mixed model) in which the outcome variable was binary (fluid responsive: yes or no) and the predictor was ΔZsys normalized to the previous phase (before the fluid challenge). We used bootstrapping (1000 replicates) from the original data at the subject level and with replacement to estimate 95% confidence intervals (CIs) of the area under the ROC curve,24 sensitivity, and specificity. We present values of sensitivity and specificity at the cutoff value chosen to maximize Youden index.

To assess the influence of the ventilatory parameters on the association between ΔZsys and SV, we used a linear mixed model, to which we tested the addition of 2 categorical variables each representing PEEP level and VT, the interaction between the 2, and the interactions of each one of them with SV.

We also explored the influence of body dimensions in the relationship between ΔZsys and SVTPT with linear mixed-effect models. We obtained the marginal R2LMM from the linear mixed-effects models according to the study by Nakagawa and Schielzeth.23 We assessed the following as potential predictors of ΔZsys: SV, weight, length, thoracic circumference, and body surface area.25

We used linear regression between ΔZsys and SVTPT to calibrate EIT in milliliters (SVEIT) from SVTPT.17 This procedure does not account for the variability in intercept and slope between subjects and was performed only to provide visual interpretation of relative changes in ΔZsys compared to changes in SVTPT. After this calibration procedure (against the reference standard), agreement between calibrated ΔZsys (SVEIT) and SVTPT was evaluated by Bland–Altman graphical analysis for repeated measurements.26

By using correlation for longitudinal data (repeated measures) to compare changes in ΔZsys and SV in relation to baseline, we estimated that a sample of 12 animals would be required considering that we would have 16 observations per animal and an expected correlation coefficient of 0.8,17 power of 0.8, and alpha of .05.

The statistical analyses were performed in R (R Core Team (2016). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL: https://www.R-project.org/.) and using software SPSS 20.0 (SPSS, Inc, Chicago, IL). P < .05 was considered significant.

This manuscript adheres to the applicable Equator guidelines.

Back to Top | Article Outline

RESULTS

Twelve female adult pigs (Agroceres) were studied. The subjects characteristics and blood volume withdrawn are presented in Table 1. A total of 204 paired measurements of SV and ΔZsys were obtained with an average of 17 measurements per pig. In 4 experiments, EIT data during ventilation with PEEP 5 cm H2O presented technical problems and were not included in the analysis.

Table 1.

Table 1.

SV and ΔZsys during each ventilatory setting through all study phases are described in Table 2. PEEP level and VT did not significantly affect the relationship between ΔZsys and SV. With that result, we did not adjust any of the other analyses for the ventilator settings. The protocol was effective in causing significant variation in SVTPT (Figure 1A) and in ΔZsys compared to the hemorrhagic phase (Figure 1B).

Table 2.

Table 2.

Figure 1.

Figure 1.

The linear association between absolute measurements of ΔZsys and SVTPT was poor (R2LMM of 0.35, P < .001). Adding information on body dimensions improved that association (R2LMM of 0.85, P < .001; Table 3). Figure 2 depicts the good linear association (R2LMM of 0.60) between ΔZsys and SVTPT indexed by weight.

Table 3.

Table 3.

Figure 2.

Figure 2.

Figure 3 shows that changes in ΔZsys had good trending ability for changes in SVTPT in relation to baseline values with concordance rates of 91.2%. The good correlation is also shown in the calibrated ΔZsys (Supplemental Digital Content 3, Figure, http://links.lww.com/AA/B849). Please be aware that we used linear regression against SVTPT to calibrate ΔZsys in milliliters (SVEIT). Consequently, the correlation we show in Supplemental Digital Content 3, Figure, http://links.lww.com/AA/B849 must be interpreted with caution. Agreement between calibrated SVEIT and SVTPT was also acceptable with negligible bias and 95% limits of agreement from −7.97 to 7.97 mL (Figure 4).

Figure 3.

Figure 3.

Figure 4.

Figure 4.

Figure 5.

Figure 5.

The ROC curve demonstrated that ΔZsys could accurately track changes in SV after a fluid challenge with an area under the curve of 0.83 (95% CI, 0.74–0.92). The results did not change significantly in a sensitivity analysis using the fluid responsiveness cutoff of 15% with an area under the curve of 0.85 (95% CI, 0.78–0.94). At the 8% cutoff for ΔZsys variation after a fluid challenge, sensitivity was 82% (95% CI, 0.76–0.89) and specificity was 75% (95% CI, 0.68–0.83) for fluid responsiveness (Figure 5A). Figure 5B shows that ΔZsys changes could discriminate responders and nonresponders to a fluid challenge. At the cutoff of 23% (dashed line), specificity was 91% (95% CI, 0.86–0.95).

Back to Top | Article Outline

DISCUSSION

We studied ΔZsys and SV in an experimental model of hemorrhagic shock. The linear association between these 2 variables was very strong within subjects, suggesting that ΔZsys can be used to track changes in SV. This was compatible with our finding of good trending ability in the 4 quadrant plot and of good discrimination to identify changes in SV after a fluid challenge. The linear association between absolute values of ΔZsys and SV was weak. Differences in body dimensions explained most of the discrepancies between individuals, such that after it was taken into account, we were able to show a good correlation between ΔZsys and absolute values of SV.

Our finding of a very strong within-experiment correlation together with an intercept that is close to 0 is compatible with efficient tracking of percent changes in SV (Figure 2 and Supplemental Digital Content 3, Figure, http://links.lww.com/AA/B849). When the aim is to assess relative changes, the scaling of ΔZsys to body dimensions becomes irrelevant. This is in agreement with the findings of Marquis et al27 who showed that anthropometric parameters did not influence impedance changes related to ventilation. Relative changes are commonly used to guide clinical decisions in the context of functional hemodynamics because static measures of preload poorly predict fluid responsiveness compared to dynamic indices.8,9 Indeed, patients are usually classified as responders or nonresponders, eg, to fluid challenges or to passive leg raising, based on whether they had more than 10%–15% increase in SV or CO after the intervention.10,19 For example, a recent systematic review demonstrated that only 50% of hemodynamic unstable patients are fluid responders and that relative changes in CO following a passive leg raising maneuver have excellent diagnostic accuracy to recognize these patients.10

We found that changes in ΔZsys had high discriminatory power in the ROC analysis (Figure 5A). This finding suggests that changes in ΔZsys, a noninvasive measure, can potentially substitute changes in SV in the assessment of fluid responsiveness. Changes in ΔZsys had high specificity and moderate sensitivity at the cutoff of 23% to identify patients who responded to a fluid challenge. This implies that changes in ΔZsys after a fluid bolus can be used to avoid unnecessary and potentially harmful additional fluid boluses.28 Alternatively, future studies could focus on the predictive ability of changes in ΔZsys after a leg raising maneuver or some other form of transiently increasing preload.

We found that changes in PEEP or VT did not influence the relationship between ΔZsys and SV (Table 2). This finding is in contrast to those of Pikkemaat et al,17 who found that ΔZsys in the heart ROI was variably affected by PEEP. In their study, it is possible that the displacement of the heart ROI in the craniocaudal axis caused by changes in PEEP modified the relative position of the heart and electrode belt. Other authors also used a cardiac ROI or used a fixed portion of the thorax ignoring whether the pixel represented heart or lung.15,29,30 In our study, a pulmonary ROI was automatically segmented by identifying pixels in which ΔZsys had a negative deflection right after the EKG–QRS. As a consequence, we imaged a cross section of the pulmonary vascular bed, thus avoiding issues related to movement of the heart in relation to the electrode belt.

Our finding that body dimensions affect the relationship between absolute values of ΔZsys and SV is supported by the relative nature of our EIT measurements. EIT outputs are difference images20,21 based on relative changes in impedance in relation to a reference, assuming a fixed shape of the thorax. The output of the reconstruction algorithm is thus expressed in percentage change in impedance in relation to the reference condition. While this approach improves the robustness of the image reconstruction, it produces outputs that are insensible to the absolute values of impedance. Therefore, an equal absolute perturbation, for instance a SV of 50 mL of blood, will produce a greater impedance change in a small than in a large thorax. We found that body dimensions were the link between the relative EIT outputs and the absolute measurements of SV. From a less technical perspective, ΔZsys has a good correlation with SV indexed to body weight (Figure 2).

Our study has some limitations. First, we only studied normal lung conditions. The relationship between SV and ΔZsys could be different in subjects with lung disease.31 Second, our hemorrhagic shock model caused extreme volume depletion, a condition seldom seen in critically ill patients. Third, we showed that we were able to track changes in systolic volume following fluid infusion, not to predict fluid responsiveness. A clinical study to evaluate the accuracy of changes in ΔZsys to predict fluid responsiveness using, for example, passive leg-raising maneuver would be welcome. Fourth, EKG gating averaged inspiration and expiration, which prevented us from assessing the effect of respiration on ΔZsys. More recent methods such as the analysis of principal components could theoretically circumvent this limitation.17,31,32 Finally, we did not test whether EIT measurements would be influenced by pneumoperitoneum or electrosurgery. EIT images are relative to a reference condition. If the reference is taken before the pneumoperitoneum, the pneumoperitoneum will cause an increase in impedance, which will be detected by the EIT, at least during the induction phase. A clinical study evaluating the effect of pneumoperitoneum and electrosurgery in EIT measurements would be welcome. Interpretation of EIT measurements should probably be restricted to periods when the electrosurgical instrument is not in use.

In conclusion, hemodynamic monitoring of SV by EIT is a promising tool. Changes in ΔZsys can reliably track relative changes in SV independent of body dimensions and could be used to optimize hemodynamics in normal lung conditions. Conversely, to produce absolute estimates of SV, the influence of body dimensions should be taken into account.

Back to Top | Article Outline

DISCLOSURES

Name: Fernando José da Silva Ramos, MD.

Contribution: This author helped acquire the data, analyze and interpret the data, and draft or revise the article for important intellectual content and final approval of data and was accountable for all aspects of the work.

Conflicts of Interest: None.

Name: André Hovnanian, MD, PhD.

Contribution: This author helped draft or revise the article for important intellectual content.

Conflicts of Interest: None.

Name: Rogério Souza, MD, PhD.

Contribution: This author helped draft or revise the article for important intellectual content.

Conflicts of Interest: None.

Name: Luciano C. P. Azevedo, MD, PhD.

Contribution: This author helped draft or revise the article for important intellectual content.

Conflicts of Interest: None.

Name: Marcelo B. P. Amato, MD, PhD.

Contribution: This author helped analyze and interpret the data and draft or revise the article for important intellectual content.

Conflicts of Interest: M. B. P. Amato received consulting fees from Timpel, Brazil.

Name: Eduardo L. V. Costa, MD, PhD.

Contribution: This author helped acquire the data, analyze and interpret the data, and draft or revise the article for important intellectual content and final approval of data and was accountable for all aspects of the work.

Conflicts of Interest: E. L. V. Costa received consulting fees from Timpel and Magnamed, both in Brazil.

This manuscript was handled by: Maxime Cannesson, MD, PhD.

Back to Top | Article Outline

REFERENCES

1. Pinsky MR. Hemodynamic evaluation and monitoring in the ICU. Chest. 2007;132:2020–2029.
2. Vincent JL, Rhodes A, Perel A, et al. Clinical review: update on hemodynamic monitoring—a consensus of 16. Crit Care. 2011;15:229.
3. Teboul JL, Saugel B, Cecconi M, et al. Less invasive hemodynamic monitoring in critically ill patients. Intensive Care Med. 2016;42:1350–1359.
4. Sandham JD, Hull RD, Brant RF, et al.; Canadian Critical Care Clinical Trials Group. A randomized, controlled trial of the use of pulmonary-artery catheters in high-risk surgical patients. N Engl J Med. 2003;348:5–14.
5. Richard C, Monnet X, Teboul JL. Pulmonary artery catheter monitoring in 2011. Curr Opin Crit Care. 2011;17:296–302.
6. Marik PE. Obituary: pulmonary artery catheter 1970 to 2013. Ann Intensive Care. 2013;3:38.
7. Rajaram SS, Desai NK, Kalra A, et al. Pulmonary artery catheter for adult patients in intensive care. Cochrane Database Syst Rev. 2013CD003408.
8. Michard F, Teboul JL. Predicting fluid responsiveness in ICU patients: a critical analysis of the evidence. Chest. 2002;121:2000–2008.
9. Pinsky MR. Understanding preload reserve using functional hemodynamic monitoring. Intensive Care Med. 2015;41:1480–1482.
10. Bentzer P, Griesdale DE, Boyd J, MacLean K, Sirounis D, Ayas NT. Will this hemodynamically unstable patient respond to a bolus of intravenous fluids? JAMA. 2016;316:1298–1309.
11. Gan TJ, Soppitt A, Maroof M, et al. Goal-directed intraoperative fluid administration reduces length of hospital stay after major surgery. Anesthesiology. 2002;97:820–826.
12. Lopes MR, Oliveira MA, Pereira VO, Lemos IP, Auler JO Jr, Michard F. Goal-directed fluid management based on pulse pressure variation monitoring during high-risk surgery: a pilot randomized controlled trial. Crit Care. 2007;11:R100.
13. Costa EL, Borges JB, Melo A, et al. Bedside estimation of recruitable alveolar collapse and hyperdistension by electrical impedance tomography. Intensive Care Med. 2009;35:1132–1137.
14. Costa EL, Chaves CN, Gomes S, et al. Real-time detection of pneumothorax using electrical impedance tomography. Crit Care Med. 2008;36:1230–1238.
15. Vonk-Noordegraaf A II, Janse A, Marcus JT, et al. Determination of stroke volume by means of electrical impedance tomography. Physiol Meas. 2000;21:285–293.
16. Maisch S, Bohm SH, Solà J, et al. Heart-lung interactions measured by electrical impedance tomography. Crit Care Med. 2011;39:2173–2176.
17. Pikkemaat R, Lundin S, Stenqvist O, Hilgers RD, Leonhardt S. Recent advances in and limitations of cardiac output monitoring by means of electrical impedance tomography. Anesth Analg. 2014;119:76–83.
18. Paul NS, Kashani H, Odedra D, Ursani A, Ray C, Rogalla P. The influence of chest wall tissue composition in determining image noise during cardiac CT. AJR Am J Roentgenol. 2011;197:1328–1334.
19. Cecconi M, Parsons AK, Rhodes A. What is a fluid challenge? Curr Opin Crit Care. 2011;17:290–295.
20. Brown BH. Electrical impedance tomography (EIT): a review. J Med Eng Technol. 2003;27:97–108.
21. Costa EL, Lima RG, Amato MB. Electrical impedance tomography. Curr Opin Crit Care. 2009;15:18–24.
22. Critchley LA, Lee A, Ho AM. A critical review of the ability of continuous cardiac output monitors to measure trends in cardiac output. Anesth Analg. 2010;111:1180–1192.
23. Nakagawa S, Schielzeth H. A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods Ecol Evol. 2013;4:133–142.
24. Liu H, Li G, Cumberland WG, Wu T. Testing statistical significance of the area under a receiving operating characteristics curve for repeated measures design with bootstrapping. J Data Sci. 2005;3:257–278.
25. Swindle MM. CRC Press; Swine in the Laboratory: Surgery, Anesthesia, Imaging, and Experimental Techniques. 2007.Boca Raton.
26. Bland JM, Altman DG. Agreement between methods of measurement with multiple observations per individual. J Biopharm Stat. 2007;17:571–582.
27. Marquis F, Coulombe N, Costa R, Gagnon H, Guardo R, Skrobik Y. Electrical impedance tomography’s correlation to lung volume is not influenced by anthropometric parameters. J Clin Monit Comput. 2006;20:201–207.
28. Wiedemann HP, Wheeler AP, Bernard GR, et al.; National Heart, Lung, and Blood Institute Acute Respiratory Distress Syndrome (ARDS) Clinical Trials Network, Comparison of two fluid-management strategies in acute lung injury. N Eng J Med. 2006;354:2564–2575.
29. Smit HJ, Vonk Noordegraaf A, Marcus JT, Boonstra A, de Vries PM, Postmus PE. Determinants of pulmonary perfusion measured by electrical impedance tomography. Eur J Appl Physiol. 2004;92:45–49.
30. Fagerberg A, Stenqvist O, Aneman A. Monitoring pulmonary perfusion by electrical impedance tomography: an evaluation in a pig model. Acta Anaesthesiol Scand. 2009;53:152–158.
31. Borges JB, Suarez-Sipmann F, Bohm SH, et al. Regional lung perfusion estimated by electrical impedance tomography in a piglet model of lung collapse. J Appl Physiol (1985). 2012;112:225–236.
32. Deibele JM, Luepschen H, Leonhardt S. Dynamic separation of pulmonary and cardiac changes in electrical impedance tomography. Physiol Meas. 2008;29:S1–S14.

Supplemental Digital Content

Back to Top | Article Outline
Copyright © 2017 International Anesthesia Research Society