Imaging the Lung in ARDS: A Primer

Abstract

Despite periodic changes in the clinical definition of ARDS, imaging of the lung remains a central component of its diagnostic identification. Several imaging modalities are available to the clinician to establish a diagnosis of the syndrome, monitor its clinical course, or assess the impact of treatment and management strategies. Each imaging modality provides unique insight into ARDS from structural and/or functional perspectives. This review will highlight several methods for lung imaging in ARDS, emphasizing basic operational and physical principles for the respiratory therapist. Advantages and disadvantages of each modality will be discussed in the context of their utility for clinical management and decision-making.

Introduction

Respiratory failure from ARDS is associated with high mortality,^1,2 and survivors may exhibit substantial long-term morbidity.³ Only modest improvements in patient outcomes have occurred since its first description by Ashbaugh et al in 1967.^4-6 Despite the variety of risk factors associated with the development of ARDS,⁷ its clinical presentation is remarkably uniform across populations, being characterized by progressive deterioration in lung function with fulminant hypoxemic respiratory failure. The management of ARDS largely focuses on supportive mechanical ventilation. Protective ventilation strategies advocate for appropriate levels of PEEP to reduce end-expiratory opening and closing,^8,9 as well as low tidal volume (V_T) or driving pressure (ΔP) to limit inspiratory overdistention.^10,11 Mechanical stresses within the lung parenchyma from repeated opening and closing of airways, as well as cyclic, intratidal overdistention, result in the release of inflammatory mediators that may further exacerbate existing injury.¹² This ventilator-induced lung injury (VILI) leads to further maldistribution of ventilation and impairments in gas exchange.¹³ In the setting of reduced V_T or ΔP, increasing breathing frequency may be the only means available to maintain acceptable CO₂ elimination and acid/base balance, although high breathing frequency may also be an independent risk factor for additional injury.¹⁴ Alternative protective ventilation strategies have also been proposed to support the failing respiratory system with ARDS, including oscillatory ventilation,¹⁵ biologically variable ventilation,¹⁶ or variants of airway pressure-release ventilation.¹⁷ However, definitive outcome measures to support their routine use are lacking.^18-23 Other therapeutic techniques that may improve outcomes in select patients include prone positioning,²⁴ neuromuscular blockade,²⁵ fluid restriction,²⁶ or extracorporeal membrane oxygenation.^27-29 Given the tremendous inter-patient variability in ARDS presentation, morphology, pathophysiology, and outcomes, a more personalized approach to mechanical ventilation is highly desired.³⁰

The exact definition of ARDS is based on clinical criteria, which are subject to revision as our understanding of its pathophysiology evolves.^31-34 The most recently published criteria appeared in 2012 as the Berlin definition, which incorporates (1) timing of acute insult to the respiratory system, (2) appearance of the lungs on chest imaging (ie, plain radiograph or computed tomography [CT]), (3) respiratory failure not explained by heart failure or volume overload, and (4) the severity of hypoxemia on minimum PEEP of 5 cm H₂O.³⁵ A modification to this definition occurred in 2016 to capture cases that may be lost in developing areas that have limited access to arterial blood gases measurements, radiographic imaging, or even mechanical ventilators.³⁶

Despite periodic changes in clinical definition of ARDS, lung imaging has consistently remained a central component of its diagnostic criteria. Several imaging modalities are available to the clinician to establish a diagnosis of ARDS, monitor its progression, or assess the impact of its treatment and management (Table 1). Each modality provides unique insight into the syndrome, from structural and/or functional perspectives. Whereas the simple chest radiograph has remained a cornerstone of ARDS diagnosis, the more recent use of lung ultrasound (US) and electrical impedance tomography (EIT) now allows for rapid, bedside assessment of lung function. Other modalities, such as positron emission tomography (PET) and magnetic resonance imaging (MRI) are limited to research studies involving small case series or pre-clinical animal models of ARDS. Nonetheless, the goal of imaging in ARDS is to assess derangements in lung aeration, recruitability, ventilation distribution, or perfusion such that efficacious and individualized treatment protocols can be devised. Central to any imaging modality in the injured lung is its ability to quantify how particular morphologic features or functional parameters respond to specified ventilation strategies. This review will highlight several methods for lung imaging in ARDS and discuss each modality roughly in order of prevalence of clinical use, not necessarily clinical utility. In contrast to many other excellent reviews on the topic,^37-46 the current article will emphasize basic operational principles and physics specifically for the respiratory therapist, discuss their respective advantages and disadvantages for clinical management and decision-making, and consider their implications for respiratory support in the patient with ARDS.

Table 1.

Advantages and Disadvantages of Different Imaging Modalities for Assessment of ARDS

Chest Radiograph

The chest radiograph has been a fundamental component of every clinical definition of ARDS since the first established criteria in 1971.³¹ Descriptions of chest radiographs in patients were reported as far back as the original 1967 report of Ashbaugh et al.⁴ Whereas the radiograph represents an anatomic image, it is not a true cross section of the chest. Rather, it is a 2-dimensional projection (or shadow), which is obtained using a radiographic source and a detector. Once the patient is appropriately positioned for imaging, the radiograph takes < 1 s to obtain. The radiation dose of a single exposure is about 0.1 mSv, similar to that received during a transoceanic flight.³⁸ The radiograph displays the contrast of superimposed anatomic structures according to their relative attenuation of radiograph as photons pass through the thorax. Accordingly, the radiograph is very sensitive for detecting pulmonary edema or patchy infiltrates typical of ARDS presentation. Although simple, the plain radiograph remains the most widely used imaging tool for establishing the diagnosis of ARDS. Given that it is highly accessible, the plain radiograph allows easy monitoring of disease progression and treatment response (Fig. 1), and bedside adjustments in mechanical ventilation are frequently based on its appearance.^47,48 Moreover, it remains an indispensable tool for verification of central venous line placement or endotracheal tube position.^49,50 The emergence of direct digital radiography also allows for ease of distribution and immediate availability.

Fig. 1.

Example chest radiographs in a male patient with ARDS, illustrating the evolution of lung injury from hospital admission (A) over the course of 3 days (B–D). From Reference 40 with permission under terms of the Creative Commons Attribution 4.0 International License.

There are some notable disadvantages of the plain chest radiograph, including anatomic superposition and geometric distortion of the projected image (especially in bedridden patients), as well as motion artifacts.⁵¹ Moreover, it has limited sensitivity and specificity for detecting pathologic abnormalities, such as distinguishing air-space infiltration versus pleural effusions.³⁹ Nonetheless, with ongoing advancements in deep learning and artificial intelligence,^52-55 the plain radiograph is expected to remain an important and widely used component of clinical diagnosis and decision support in ARDS management.

Computed Tomography

CT is another radiographic imaging modality, albeit with enhanced capability for distinguishing pathologic features associated with ARDS, with high spatial resolution and views in multiple anatomic sections. Rather than delivering a quick burst of x-rays that penetrates the patient, the CT scanner relies on a rotating emitting source, which projects x-rays through the patient and onto an opposing detector array (Fig. 2A). This technology has evolved considerably over the past 40 years, with radiation exposure now down to not much more than a pair of plain chest radiographs and image acquisition requiring just several seconds or less.⁵⁶ With volumetric CT imaging, the radiographic source rotates at a fixed frequency around a table that moves at a constant speed in a single direction. The radiographic tube then traces out a helical path along the recumbent patient’s cephalocaudal (or vice versa) body axis (Fig. 2B). Based on the resulting projection data captured by the radiographic detectors juxtaposed to the radiographic source, computer software reconstructs the cross-sectional images using mathematical interpolations, yielding a “stack” of image slices, each with a specified thickness. The result will be a 3-dimenional anatomic map based on gray scale density.

Fig. 2.

A: Illustration of radiographic tube and detector array rotating around subject with the computed tomography gantry. B: As the radiographic tube and detector array rotates, the table moves through the gantry with a constant velocity, resulting in a helical path of projection data around the subject. The helical projection data can then be reconstructed into transverse planes using interpolation, thus yielding a stack of images. From Reference 117 with permission. Copyright © 2002 Wolters Kluwer Health, Inc. All rights reserved.

The level of “grayness” is precisely defined in the CT image, according to the Hounsfield scale. Hounsfield units (HUs) are discrete integers corresponding to the level of radiographic attenuation, with a HU of zero corresponding to distilled water and −1,000 corresponding to pure air. Thus, the grayness of every discrete element within the lung image, or voxel, is assumed to correspond to a linear combination of air and water.^57-59 The voxel is a 3-dimensional element with a height corresponding to the slice thickness (Fig. 3A). With current CT scanners, one voxel is about ≤ 1 mm³, about the size of 160–170 alveoli.³⁸ This size limitation has implications for spatial resolution. For example, 2 different voxels may have the same average air and water content and thus the same level of CT grayness (or density), despite containing very different alveolar structures (Fig. 3B). Moreover, CT images are often “windowed” onto a video monitor depending on the organ system of interest, since the human eye is not capable of discerning 1,000 discrete levels of grayness and the monitor does not accommodate that many levels. A display window will have a width, corresponding to the range of HU translated onto the monitor’s display range, typically 256 levels. Above the range, all the voxels appear white; and below the range, they all appear black. A typical width for the lung is about 1,500 HU. The window level is the location on the Hounsfield scale where the window is centered, which is typically −600 HU when viewing the lung (Fig. 4). The selected window and level can significantly change the visual assessment of the organ structures of interest, and thus alter the interpretation. Accordingly, it is critical to standardize the display parameters.

Fig. 3.

A: Example of a single 1 mm³ voxel from a computed tomography (CT) image in transverse (ie, axial) cross section of the thorax. At this spatial resolution, one voxel contains about 170 alveoli. From Reference 117 with permission. Copyright © 2002 Wolters Kluwer Health, Inc. All rights reserved. B: Given this spatial resolution, 2 different voxels may have the same level of CT grayness (or density) since they have the same air and water contents on average, despite very different alveolar structure. From Reference 38 with permission. Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. Anesthesiology is an official journal of the American Society of Anesthesiologists. Readers are encouraged to read the entire article for correct content at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6692186/. The original authors, editors, and the American Society of Anesthesiologists are not responsible for errors or omissions in adaptations.

Fig. 4.

Illustration of computed tomography (CT) window (W) and level values. The level (L) corresponds to the center of the window, with a narrow window producing a high-contrast image. CT densities below P₁ will appear black, while those above P₂ will appear white. Only CT densities between P₁ and P₂ will vary according to radiographic attenuation. CT = computed tomography. From Reference 117 with permission. Copyright © 2002 Wolters Kluwer Health, Inc. All rights reserved.

There are several ways by which a CT image can be interpreted. The most common (and familiar) would be a qualitative interpretation, which is how a clinical radiologist would read the image. Notable features would be the presence and size of pleural effusions, ground glass opacities, consolidation, placement of lines and tubes, etc. Qualitative interpretations are, of course, subjective and thus may exhibit intra-observer variability. Since the initial use of the modality in the 1980s,^60,61 CT imaging has demonstrated that the pathology of ARDS is much more heterogeneous than plain chest radiographs would indicate, (Fig. 5) with more ventral regions in the supine patient having better aeration and more dorsal regions being consolidated.

Fig. 5.

Comparison of (A) a plain chest radiograph with (B) a computed tomography (CT) scan in ARDS. Images were obtained in the same patient, close together in time. Note the more specific structural information obtained with the CT scan. Whereas both images demonstrate disuse infiltrates, the CT image indicates that these are located primarily in the dorsal, gravity-dependent regions (indicated by the black arrow), with the ventral regions of the lung having greater aeration (indicated by green arrow). From Reference 38 with permission. Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. Anesthesiology is an official journal of the American Society of Anesthesiologists. Readers are encouraged to read the entire article for correct content at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6692186/. The original authors, editors, and the American Society of Anesthesiologists are not responsible for errors or omissions in adaptations.

Quantitative interpretations, by definition, are much more objective⁶² and have evolved over several decades.⁵⁶ These interpretations rely on tools such as image segmentation,⁶³ density analysis,⁶⁴ and parenchymal texture analysis.⁶⁵ Before a quantitative analysis can begin, the lungs must first be segmented from the surrounding chest-wall structures and mediastinum. In the healthy lung, this is relatively easy using automated density threshold techniques,⁶⁶ although manual correction may also be required.⁶⁷ Automated segmentation becomes much more difficult for the lung with ARDS since lung consolidation may be poorly distinguished from other surrounding structures, such as paraspinous muscles, the mediastinum, or pleural effusions. Recent studies have demonstrated the potential of machine learning algorithms for segmenting injured lungs,^68,69 especially if appropriate training data are available.⁷⁰ Once the parenchyma has been segmented, the statistical properties of Hounsfield density within the lung can be interpreted with a histogram (Fig. 6). Moreover, the level of gray scale can also be compartmentalized based on arbitrary, but well-accepted, Hounsfield ranges: voxels with densities < −900 HU are considered hyper-aerated; those between −900 HU to −500 HU are normally aerated; voxels −500 to −100 HU are poorly aerated, and voxels > −100 HU are non-aerated.

Fig. 6.

Example density histograms from a small cohort of patients with ARDS, plotting a distribution of the lung volume as a function of Hounsfield density for zero airway pressure, as well as lower and higher PEEPs. For clarity, error bars are omitted. Also shown are compartmentalized aeration ranges, corresponding to hyper-aerated, normally aerated, poorly aerated, and non-aerated regions. Note that as the level of airway pressure increases a greater proportion of total lung volume occurs in these more aerated regions, while conversely the amount of non-aerated lung decreases. From Reference 132 with permission. Copyright © 1999 American Thoracic Society. All rights reserved. The American Journal of Respiratory and Critical Care Medicine is an official journal of the American Thoracic Society. Readers are encouraged to read the entire article for the correct context at https://doi.org/10.1164/ajrccm.159.5.9805112. The original authors, editors, and the American Thoracic Society are not responsible for errors or omissions or adaptations.

Finally, functional interpretation goes beyond solely static spatial and structural information and analyzes how the lung functions under operational conditions, such as during spontaneous breathing or mechanical ventilation. Included in functional interpretations are (1) dynamic reconstruction to monitor intratidal variations in lung recruitment and aeration;⁷¹ (2) image registration to assess regional parenchymal deformation and strain during inflation;^67,72 and (3) dual-energy CT with inhaled and/or intravenous contrast enhancement to quantify the distribution of ventilation and perfusion, respectively.^37,62,73

CT imaging possesses a number of advantages in comparison to other imaging modalities. It remains the accepted standard by which lung aeration is assessed.^74,75 Moreover, it has higher sensitivity and specificity for detecting pathologic and structural abnormalities compared to other modes of lung imaging. When obtained with inhaled⁷⁶ or intravenous⁷³ contrast enhancement of the air spaces or vasculature, respectively, as well as multispectral radiographic energy, CT provides insight into lung recruitment and consolidation, regional distribution of aeration and perfusion, and the anatomic location of their respective intratidal variability. Finally, CT imaging may also be useful for distinguishing different ARDS phenotypes and/or stages of lung injury.^40,77,78

There are notable disadvantages with CT imaging, however, especially with regard to radiation dose: A single CT scan of the thorax exposes a patient to about 7 mSv of radiation, about the same exposure as 2 cumulative years of background radiation. This has limited the routine use of CT imaging for rapid adjustments in ventilator settings, such as PEEP, V_T, or ΔP. With the advancement of computational capabilities along with detector and radiographic tube technology, radiation exposures < 1 mSv can be achieved. Spatial resolution can reach 0.2–0.3 mm with the introduction of photon-counting CT, whereby each photon passing through the body is captured digitally and binned into individual energy ranges.^79-81 There are also risks associated with the administration of contrast, especially for intravenous delivery of iodinated media that may result in allergic reactions and nephrotoxicity.⁸² With regard to quantitative or functional CT image analyses, long post-processing times have made some techniques computationally prohibitive in the clinical arena. However, these barriers are fading. Outcome benefits of CT imaging are also not established, with a recent clinical trial of personalized ventilation based on this modality found no improvement in mortality compared to a traditional low V_T approach based on ideal body weight.⁸³ Finally, there are risks associated with transporting an intubated and ventilated patient from an ICU to an imaging suite, with adverse events reported to be between 4–68%, depending on incident severity and patient condition.^39,84 However, the emergence and use of portable CT scanners during the peak of the COVID-19 pandemic may make CT imaging technology more available for bedside implementation.⁸⁵

Lung Ultrasound

In contrast to plain radiographs or CT imaging, US involves no exposure to ionizing radiation. With US, sound waves are generated from a piezoelectric material in a handheld probe at frequencies of 1–12 MHz.⁸⁶ These sound waves will penetrate the tissues with a depth that depends on frequency. Portions of the sound energy will be reflected back to the US probe, usually at the interface between tissues with different acoustic impedances. Reconstructed images will be based on time of flight, as well as the intensity of the reflection.³⁸ In contrast to solid organs such as the kidneys, liver, or heart, US images from the lung will be artifact-based since an air-filled lung cannot be directly visualized using sound waves. Thus, what is observed as a US image arises from the reconstruction algorithms that generate erroneous, non-anatomic information. This does not mean that these artifacts have no clinical utility. Indeed, such artifacts may be broadly categorized into groups based on aeration patterns (Fig. 7).³⁸ For example, so-called horizontal A-lines are regularly spaced reverberation artifacts of the pleural surface, indicating a relatively well-aerated region of the lung. If the lung region becomes de-aerated, such horizontal A-lines will be absorbed by vertical B-lines (ie, “comet-tails”), the number of which will correspond to the disease severity and de-aeration, whether this be due to edema, fibrosis, or inflammation.⁸⁶ Consolidation occurs when there is complete absence of alveolar air, and in this case, one may visualize an actual anatomic image of the lung tissue.

Fig. 7.

Examples of lung ultrasound images from a normal lung (A), lungs with decreased aeration due to interstitial syndrome (B), and lungs with complete consolidation (C). In addition to shadows from the underlying ribs, the healthy lung in (A) also shows a stripe corresponding to the pleural surface, with A-line reverberation artifacts. From Reference 38 with permission. Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. Anesthesiology is an official journal of the American Society of Anesthesiologists. Readers are encouraged to read the entire article for correct content at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6692186/. The original authors, editors, and the American Society of Anesthesiologists are not responsible for errors or omissions in adaptations.

In addition to providing an image without exposure to ionizing radiation, additional advantages of US include its low cost; portability; and ability to yield immediate, bedside assessment of lung aeration.⁸⁷ Lung US also has high sensitivity and specificity for detecting structural abnormalities, such as pneumothoraces, pleural effusions, or pulmonary contusions.^40,86 Moreover, US may allow for the detection of improved aeration (ie, lung recruitment) in real time, thus potentially allowing for bedside PEEP titration. Disadvantages include the interpretation of a very non-intuitive, artifact-based image for aerated lung regions; the lack of specificity with regard to the cause of de-aeration; and its inability to detect parenchymal overinflation (an important component of VILI).⁴⁴ Lung US is also quite time-consuming, taking up to 30 min for a highly methodical, comprehensive examination within 6, 8, or even 12 zones of the chest due to limited field of view, low sound wave penetration depth, and rib shadows. Finally, operator proficiency requires substantial training, with an estimated 25 supervised examinations to achieve minimal competence.⁸⁸

Electrical Impedance Tomography

EIT shares some similarities with radiographic CT in that the image presented is a tomographic reconstruction. However, rather than discerning anatomic structures based on differences in radiographic absorption and penetration, EIT uses alternating electrical currents to detect regional changes in electrical impedance, that is, the complex ratio of voltage to current. To obtain EIT measurements, a belt containing 16 or 32 electrodes is placed around the rib cage (usually at the fifth intercostal space), forming a ring around thoracic structures in the transverse plane.⁸⁹ Low-dose alternating microcurrents are then injected from pairs of adjacent electrodes, and corresponding voltages are sensed by the remaining electrodes (Fig. 8). This cycle is repeated using different electrode pairs, and the voltage responses are traced throughout the whole belt. The electrical impedance of biological tissues will vary depending on their composition. For example, tissues with high water and electrolyte content will have low impedance, while air, fat, and bone will have high impedance. The image that is reconstructed thus corresponds to the changes in the spatial distribution of aeration during a breath. EIT images are usually compared to a reference image at a baseline end-expiratory lung volume, such that the resulting “difference” images correspond to spatially distributed changes in aeration.⁹⁰ EIT images are thus not anatomic images per se but rather functional images that correspond to changes in the air content of the lungs. Accordingly, EIT allows one to track intratidal variations in regional impedance (Fig. 9).

Fig. 8.

Determination of regional electrical impedance in the thorax using electrical impedance tomography. In this example, alternating electrical current (I₁) is injected from electrodes 1 and 16, and the corresponding voltage responses are sensed by electrodes 2–15. This cycle is repeated using different electrode pairs (ie, current injection between electrodes 1 and 2, then 2 and 3, etc) until the voltage responses are traced throughout the whole belt. From Reference 38 with permission. Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. Anesthesiology is an official journal of the American Society of Anesthesiologists. Readers are encouraged to read the entire article for correct content at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6692186/. The original authors, editors, and the American Society of Anesthesiologists are not responsible for errors or omissions in adaptations.

Fig. 9.

A: Comparison between a computed tomography (CT) image (upper panel) and an electrical impedance tomography (EIT) image (lower panel). The EIT image depicts regional changes in electrical impedance according to a color scale, with red indicating maximal tidal impedance change and blue indicating no change. B: Corresponding time tracings of regional electrical impedance changes over the course of 1 min. Note that the right dorsal region of the lung is electrically silent in the EIT image, consistent with the complete consolidation of the CT mage in A. From Reference 39 with permission under terms of the Creative Commons Attribution 4.0 International License.

A major advantage of EIT is that it provides a radiation-free, bedside assessment of aeration. Moreover, it achieves this with very high temporal resolution (ie, 25–60 frames/s) and thus may even be used to quantify ventilation distribution, even during high-frequency oscillatory ventilation.^91,92 Moreover, EIT indices exist for estimating intratidal de-recruitment and overdistention,^93-96 thus theoretically allowing for personalized adjustments in PEEP, V_T, or ΔP.^97-99 Finally, EIT may also allow for assessment of lung perfusion, using pulsatility or hypertonic saline for contrast enhancement.^97,100,101

Disadvantages of EIT include its reliance on a belted electrode array, resulting in only a single, 2-dimensional cross section of about 5–10 cm thickness. However, 3-dimensional approaches using multiple transverse slices have been investigated.^102-105 Given the mathematical difficulty of converting electrical impedance measurements at the arbitrary boundary of an object into spatial distributions of electrical impedance within a transverse plane,¹⁰⁶ EIT yields images with very low spatial resolution (about 800 pixels per slice, on the order of centimeters per pixel in adult patients), which can only be modestly improved by increasing the number of electrodes.¹⁰⁷ As a functional imaging technique, EIT has clinically utility for detecting sudden losses in aeration (eg, pneumothorax or lung collapse). However, it lacks specificity for determining the etiology for absent ventilation. Moreover, EIT may not be suitable for patients with chest trauma or recent thoracic surgery. Despite the potential for personalization of mechanical ventilation with EIT, there has yet to be any proven outcome benefit established in randomized clinical trials.^108,109

Positron Emission Tomography

PET is a functional imagining modality based on the detection of a radioactive tracer on a biologically active molecule.¹¹⁰ Such a tracer may be delivered as an inspired gas or intravenous injection and results in the emission of positrons, which interact with electrons in local tissues and result in the emission of detectable gamma rays.³⁸ Depending on the specific tracer and its route of delivery, PET imaging can be used to quantify regional ventilation, perfusion, extravascular lung water, or inflammation. Moreover, the 3-dimensional locations of biochemical and inflammatory process can be followed over time.³⁸ One of the most commonly used radiolabeled molecules for PET is [¹⁸F]-fluoro-2-deoxy-D-glucose (ie, [¹⁸F]FDG), a glucose analog that allows for the visualization of blood flow, as well as metabolic and biochemical activities in healthy and diseased lung tissues. Since activated neutrophils consume relatively high quantities of glucose during anerobic glycolysis, [¹⁸F]FDG can be used to visualize inflammatory cellular metabolic activity, especially in the setting of ARDS and VILI (Fig. 10).^110,111 Thus, mechanical processes such as intratidal overdistention and de-recruitment can also be assess with [¹⁸F]FDG uptake.^112,113 Inhaled labeled nitrogen (ie,¹³N₂) can be used to assess regional aeration, recruitment, or ventilation, while its intravenous infusion can also yield information about regional perfusion and ventilation/perfusion matching.¹¹⁴ Other radiolabeled tracers, such as H₂¹⁵O, ¹¹CO, or C¹⁵O, can also distinguish intravascular from extravascular lung water.^115,116 Despite providing tremendous insight into the pathophysiology of ARDS and VILI, PET remains limited to small case series or pre-clinical animal models. Given the cost, risks of patient transport, exam duration, and radiation exposure, the clinical utility of PET for ARDS management remains to be established.⁴¹

Fig. 10.

Cross-registered computed tomography and [¹⁸F]FDG positron emission tomography (PET) images from a patient with ARDS. The PET image represents average [¹⁸F]FDG concentration during the last 20 min of acquisition (from 37–57 min since administration). Color scale represents radioactivity concentration in units of kBq/mL. Note that [¹⁸F]FDG uptake is high in normally aerated regions (square 1) but is lower in the dorsal, non-aerated regions (square 2). From Reference 111 with permission. Copyright © 2009 Wolters Kluwer Health, Inc. Permission conveyed through Copyright Clearance Center, Inc.

Magnetic Resonance Imaging

MRI is a modality that constructs anatomic images using signals generated by the rotation of protons in a strong magnetic field after a radio frequency field is toggled on and off. The image is constructed from the signals of these protons as they return to their original state after the radio frequency field is turned off. Two time constants are associated with this relaxation: T1, corresponding to the time for the protons to return to equilibrium; and T2, corresponding to the time for the protons to go out of phase with each other.¹¹⁷ Whereas MRI has been a revolutionary modality for the imaging of solid soft tissues such as the brain, liver, or musculature, it has very limited clinical applications for imaging the lungs in patients with ARDS, due to long acquisition times, low proton density in aerated lung regions, as well as loss of contrast at air/tissue interfaces due to magnetic effects.^118-120 Two-dimensional MRI acquisition does allow for faster acquisition times compared to 3 dimensions,^121,122 although this yields only a single cross section of the thoracic cavity. Ongoing advancements in MRI technology and image processing have improved acquisition capabilities for clinical support.^123-126 Ultra-short echo time sequences may have utility for discerning parenchymal tissue structure,¹²⁷ and magnetic resonance elastography may also be used to estimate shear stiffness of the lung tissues.^128,129 Hyperpolarized noble gases (ie, ³He or ¹²⁹Xe) allow for assessments of regional gas distribution, air-space dimensions via the apparent diffusion coefficient, as well as regional gas uptake in injured lungs.^38,130 The physics of MRI are varied and complicated, but the acquired images may eventually rival those of radiographic CT in spatial and temporal resolution of lung structure and function, thus allowing for quantitative discernment of the distributed, heterogeneous pattern of lung injury without harmful radiation exposure.¹³

Summary

This review has discussed various lung imaging modalities in the context of clinical usefulness for the respiratory therapist managing the patient with ARDS. The goal of any imaging modality should be to provide a mechanistic basis for tailoring mechanical ventilation in individual patients. Thus, its utility should depend on its ability to provide reliable assessment of aeration, gas transport, perfusion, or the distribution of mechanical properties in the injured lung. Moreover, it should do so with appropriate spatial and temporal resolution such that it may complement respiratory mechanical information obtained from pressure and flow measurements at the airway opening.¹³¹ Some modalities remain impractical for routine clinical use based on cost, scarcity of contrast and tracer agents, bedside implementation, exposure to ionizing radiation, and computational overhead associated with image processing. Nonetheless, they provide unique insight into pathophysiologic mechanisms of ARDS, as well as information that may validate more clinically practical approaches to personalize mechanical ventilation. Ongoing advancements in detector technology, computational hardware, and image reconstruction algorithms continue to improve image quality, spatial and temporal resolution, as well as portability. Additionally, the development of machine learning algorithms may yield automated feature extractions of lung images, with reduced computational costs. Future research in lung imaging should yield results that are translatable and testable in human clinical trials, with potential to reduce morbidity and mortality associated with ARDS.