Search

Segmentation of lungs with acute respiratory distress syndrome (ARDS) is a challenging task due to diffuse opacification in dependent regions which results in little to no contrast at the lung boundary. For segmentation of severely injured lungs, local intensity and texture information, as well as global contextual information, are important factors for consistent inclusion of intrapulmonary structures. In this study, we propose a deep learning framework which uses a novel multi-resolution convolutional neural network (ConvNet) for automated segmentation of lungs in multiple mammalian species with injury models similar to ARDS. The multi-resolution model eliminates the need to tradeoff between high-resolution and global context by using a cascade of low-resolution to high-resolution networks. Transfer learning is used to accommodate the limited number of training datasets. The model was initially pre-trained on human CT images, and subsequently fine-tuned on canine, porcine, and ovine CT images with lung injuries similar to ARDS. The multi-resolution model was compared to both high-resolution and low-resolution networks alone. The multi-resolution model outperformed both the low- and high-resolution models, achieving an overall mean Jacaard index of 0.963 ± 0.025 compared to 0.919 ± 0.027 and 0.950 ± 0.036, respectively, for the animal dataset (N=287). The multi-resolution model achieves an overall average symmetric surface distance of 0.438 ± 0.315 mm, compared to 0.971 ± 0.368 mm and 0.657 ± 0.519 mm for the low-resolution and high-resolution models, respectively. We conclude that the multi-resolution model produces accurate segmentations in severely injured lungs, which is attributed to the inclusion of both local and global features., (Copyright © 2019. Published by Elsevier B.V.)

Background: Acute respiratory distress syndrome (ARDS) is an inflammatory condition comprising diffuse lung edema and alveolar damage. ARDS frequently results from regional injury mechanisms. However, it is unknown whether detectable inflammation precedes lung edema and opacification and whether topographically differential gene expression consistent with heterogeneous injury occurs in early ARDS. The authors aimed to determine the temporal relationship between pulmonary metabolic activation and density in a large animal model of early ARDS and to assess gene expression in differentially activated regions., Methods: The authors produced ARDS in sheep with intravenous lipopolysaccharide (10 ng ⋅ kg ⋅ h) and mechanical ventilation for 20 h. Using positron emission tomography, the authors assessed regional cellular metabolic activation with 2-deoxy-2-[(18)F]fluoro-D-glucose, perfusion and ventilation with NN-saline, and aeration using transmission scans. Species-specific microarray technology was used to assess regional gene expression., Results: Metabolic activation preceded detectable increases in lung density (as required for clinical diagnosis) and correlated with subsequent histologic injury, suggesting its predictive value for severity of disease progression. Local time courses of metabolic activation varied, with highly perfused and less aerated dependent lung regions activated earlier than nondependent regions. These regions of distinct metabolic trajectories demonstrated differential gene expression for known and potential novel candidates for ARDS pathogenesis., Conclusions: Heterogeneous lung metabolic activation precedes increases in lung density in the development of ARDS due to endotoxemia and mechanical ventilation. Local differential gene expression occurs in these early stages and reveals molecular pathways relevant to ARDS biology and of potential use as treatment targets.

Purpose: Imaging has become increasingly important across medical specialties for diagnostic, monitoring, and investigative purposes in acute respiratory distress syndrome (ARDS)., Methods: This review addresses the use of imaging techniques for the diagnosis and management of ARDS as well as gaining knowledge about its pathogenesis and pathophysiology. The techniques described in this article are computed tomography, positron emission tomography, and two easily accessible imaging techniques available at the bedside-ultrasound and electrical impedance tomography (EIT)., Results: The use of computed tomography has provided new insights into ARDS pathophysiology, demonstrating that ARDS does not homogeneously affect the lung parenchyma and that lung injury severity is widely distributed in the ARDS population. Positron emission tomography is a functional imaging technique whose value resides in adding incremental insights to morphological imaging. It can quantify regional perfusion, ventilation, aeration, lung vascular permeability, edema, and inflammation. Lung ultrasound and EIT are radiation-free, noninvasive tools available at the bedside. Lung ultrasound can provide useful information on ARDS diagnosis when x-rays or CT scan are not available. EIT is a useful tool to monitor lung ventilation and to assess the regional distribution of perfusion., Conclusions: The future of imaging in critical care will probably develop in two main directions: easily accessible imaging techniques that can be used at the bedside and sophisticated imaging methods that will be used to aid in difficult diagnostic cases or to advance our understanding of the pathogenesis and pathophysiology of an array of critical illnesses.

Unlabelled: PET with (18)F-FDG allows for noninvasive assessment of regional lung metabolism reflective of neutrophilic inflammation. This study aimed at determining during early acute lung injury whether local (18)F-FDG phosphorylation rate and volume of distribution were sensitive to the initial regional inflammatory response and whether they depended on the mechanism of injury: endotoxemia and surfactant depletion., Methods: Twelve sheep underwent homogeneous unilateral surfactant depletion (alveolar lavage) and were mechanically ventilated for 4 h (positive end-expiratory pressure, 10 cm H2O; plateau pressure, 30 cm H2O) while receiving intravenous endotoxin (lipopolysaccharide-positive [LPS+] group; n = 6) or not (lipopolysaccharide-negative group; n = 6). (18)F-FDG PET emission scans were then acquired. (18)F-FDG phosphorylation rate and distribution volume were calculated with a 4-compartment model. Lung tissue expression of inflammatory cytokines was measured using real-time quantitative reverse transcription polymerase chain reaction., Results: (18)F-FDG uptake increased in LPS+ (P = 0.012) and in surfactant-depleted sheep (P < 0.001). These increases were topographically heterogeneous, predominantly in dependent lung regions, and without interaction between alveolar lavage and LPS. The increase of (18)F-FDG uptake in the LPS+ group was related both to increases in the (18)F-FDG phosphorylation rate (P < 0.05) and to distribution volume (P < 0.01). (18)F-FDG distribution volume increased with infiltrating neutrophils (P < 0.001) and phosphorylation rate with the regional expression of IL-1β (P = 0.026), IL-8 (P = 0.011), and IL-10 (P = 0.023)., Conclusion: Noninvasive (18)F-FDG PET-derived parameters represent histologic and gene expression markers of early lung injury. Pulmonary metabolism assessed with (18)F-FDG PET depends on the mechanism of injury and appears to be additive for endotoxemia and surfactant depletion. (18)F-FDG PET may be a valuable imaging biomarker of early lung injury., (© 2014 by the Society of Nuclear Medicine and Molecular Imaging, Inc.)

Objective: Neutrophilic inflammation plays a key role in the pathogenesis of acute respiratory distress syndrome (ARDS) and acute lung injury (ALI). Positron emission tomography (PET) with [F]-fluoro-2-deoxy-D-glucose (FDG) can be used to image cellular metabolism that, during lung inflammatory processes, likely reflects neutrophils activity. The aim of this study was to assess the magnitude and regional distribution of inflammatory metabolic activity in the lungs of patients with ALI/ARDS by PET with FDG., Design: Prospective clinical investigation., Patients: Ten patients with ALI/ARDS; four spontaneously breathing and two mechanically ventilated subjects, without known lung disease, served as controls., Interventions: In each individual we performed an FDG PET/computed tomography of the thorax., Measurements and Main Results: FDG cellular influx rate constant (Ki) was computed for the imaged lung field and for regions of interest, grouping voxels with similar density. In all patients with ALI/ARDS, Ki was higher than in controls, also after accounting for the increased lung density. Ki values differed greatly among patients, but in all patients Ki of the normally aerated regions was much higher (2- to 24-fold) than in controls. Whereas in some patients the highest Ki values corresponded to regions with the lowest aeration, in others these regions had lower Ki than normally and mildly hypoaerated regions., Conclusion: In patients with ALI/ARDS, undergoing mechanical ventilation since days, the metabolic activity of the lungs is markedly increased across the entire lung density spectrum. The intensity of this activation and its regional distribution, however, vary widely within and between patients.

Rationale and Objectives: Dynamic positron emission tomographic imaging of the radiotracer 2-deoxy-2-[(18)F]fluoro-D-glucose ((18)F-FDG) is increasingly used to assess metabolic activity of lung inflammatory cells. To analyze the kinetics of (18)F-FDG in brain and tumor tissues, the Sokoloff model has been typically used. In the lungs, however, a high blood-to-parenchymal volume ratio and (18)F-FDG distribution in edematous injured tissue could require a modified model to properly describe (18)F-FDG kinetics., Materials and Methods: We developed and validated a new model of lung (18)F-FDG kinetics that includes an extravascular/noncellular compartment in addition to blood and (18)F-FDG precursor pools for phosphorylation. Parameters obtained from this model were compared with those obtained using the Sokoloff model. We analyzed dynamic PET data from 15 sheep with smoke or ventilator-induced lung injury., Results: In the majority of injured lungs, the new model provided better fit to the data than the Sokoloff model. Rate of pulmonary (18)F-FDG net uptake and distribution volume in the precursor pool for phosphorylation correlated between the two models (R(2)=0.98, 0.78), but were overestimated with the Sokoloff model by 17% (P< .05) and 16% (P< .0005) compared to the new one. The range of the extravascular/noncellular (18)F-FDG distribution volumes was up to 13% and 49% of lung tissue volume in smoke- and ventilator-induced lung injury, respectively., Conclusion: The lung-specific model predicted (18)F-FDG kinetics during acute lung injury more accurately than the Sokoloff model and may provide new insights in the pathophysiology of lung injury.

Rationale: In a pulmonary process characterized by spatially heterogeneous loss of aeration, the impairment of gas exchange is expected to depend on the regional distribution of perfusion relative to that of aeration., Objectives: To investigate how regional aeration, shunt, and perfusion are interrelated at different levels of end-expiratory pressure and how their interplay relates to global shunt fraction in acute lung injury., Methods: Regional shunt and perfusion were assessed by imaging with positron emission tomography the pulmonary kinetics of [(13)N]nitrogen infused in saline solution in five sheep after lung lavage. The lung field was divided in six horizontal regions., Measurements and Main Results: Each animal showed an inverse relation between regional shunt (Fs) and gas (Fg) fractions: Fs = -m . Fg + Fs(0). This relation was similar among animals (m = 1.25 +/- 0.14, Fs(0) = 0.75 +/- 0.15) and invariant with end-expiratory pressure, despite lack of correlation between global shunt and gas fractions and large interanimal variability in global shunt fraction. When this relation was used to estimate global shunt fraction as a perfusion-weighted average of the estimates of regional shunt fraction derived from regional gas fraction, 72% of the interanimal variability in global shunt fraction could be explained., Conclusions: Despite large interanimal variability in global shunt fraction, there was a consistent inverse relation between regional shunt and gas fractions, independent of end-expiratory pressure. Most of the interanimal variability in global shunt fraction could be explained by the combined effect of this relation and the distribution of perfusion on regional shunt, rather than by differences in global aeration.

Background: Alveolar overdistension and repetitive derecruitment-recruitment contribute to ventilator-induced lung injury (VILI). The authors investigated (1) whether inflammatory cell activation due to VILI was assessable by positron emission tomography and (2) whether cell activation due to dynamic overdistension alone was detectable when other manifestations of VILI were not yet evident., Methods: The authors assessed cellular metabolic activity with [(18)F]fluorodeoxyglucose and regional gas exchange with [(13)N]nitrogen. In 12 sheep, the left ("test") lung was overdistended with end-inspiratory pressure of 50 cm H(2)O for 90 min, while end-expiratory derecruitment of this lung was either promoted with end-expiratory pressure of -10 cm H(2)O in 6 of these sheep (negative end-expiratory pressure [NEEP] group) or prevented with +10 cm H(2)O in the other 6 (positive end-expiratory pressure [PEEP] group) to isolate the effect of overdistension. The right ("control") lung was protected from VILI., Results: Aeration decreased and shunt fraction increased in the test lung of the NEEP group. [(18)F]fluorodeoxyglucose uptake of this lung was higher than that of the control lung and of the test lung of the PEEP group, and correlated with neutrophil count. When normalized by tissue fraction to account for increased aeration of the test lung in the PEEP group, [(18)F]fluorodeoxyglucose uptake was elevated also in this group, despite the fact that gas exchange had not yet deteriorated after 90 min of overdistension alone., Conclusion: The authors could detect regional neutrophil activation in VILI even when end-expiratory derecruitment was prevented and impairment of gas exchange was not evident. Concomitant end-expiratory derecruitment converted this activation into profound inflammation with decreased aeration and regional shunting.

Regional pulmonary perfusion and ventilation can be assessed by imaging, with positron emission tomography (PET), the pulmonary kinetics of [13N]nitrogen (13N2). Because of its low solubility in blood and tissues, 13N2 infused intravenously in saline solution evolves into the alveolar airspace at first pass, where it accumulates in proportion to regional perfusion during a short apnea. In contrast, infused 13N2 is not retained in non-aerated regions, which do not exchange gas. Robust estimates of regional perfusion and shunt are obtained by modeling the pulmonary kinetics of 13N2 infused as a bolus during a short apnea. Regional ventilation is measured by modeling the washout of 13N2 after breathing is resumed. Regional gas content and dead space ventilation can be measured with inhalation of 13N2. Application of this novel functional imaging technique can further the understanding of the pathophysiology of a variety of pulmonary processes. This review briefly describes the methodological aspects of PET imaging of regional perfusion and ventilation and then focuses on insights in the pathophysiology of acute lung injury and asthma that have been gained by imaging the pulmonary kinetics of 13N2 with PET.

Rationale: The prone position is used to improve gas exchange in patients with acute respiratory distress syndrome. However, the regional mechanism by which the prone position improves gas exchange in acutely injured lungs is still incompletely defined., Methods: We used positron emission tomography imaging of [(13)N]nitrogen to assess the regional distribution of pulmonary shunt, aeration, perfusion, and ventilation in seven surfactant-depleted sheep in supine and prone positions., Results: In the supine position, the dorsal lung regions had a high shunt fraction, high perfusion, and poor aeration. The prone position was associated with an increase in lung gas content and with a more uniform distribution of aeration, as the increase in aeration in dorsal lung regions was not offset by loss of aeration in ventral regions. Consequently, the shunt fraction decreased in dorsal regions in the prone position without a concomitant impairment of gas exchange in ventral regions, thus leading to a significant increase in the fraction of pulmonary perfusion participating in gas exchange. In addition, the vertical distribution of specific alveolar ventilation became more uniform in the prone position. A biphasic relation between regional shunt fraction and gas fraction showed low shunt for values of gas fraction higher than a threshold, and a steep linear increase in shunt for lower values of gas fraction., Conclusion: In a surfactant-deficient model of lung injury, the prone position improved gas exchange by restoring aeration and decreasing shunt while preserving perfusion in dorsal lung regions, and by making the distribution of ventilation more uniform.

Background: Sustained lung inflations (recruitment maneuvers [RMs]) are occasionally used during mechanical ventilation of patients with acute lung injury to restore aeration to atelectatic alveoli. However, RMs do not improve, and may even worsen, gas exchange in a fraction of these patients. In this study, the authors sought to determine the mechanism by which an RM can impair gas exchange in acute lung injury., Methods: The authors selected a model of acute lung injury that was unlikely to exhibit sustained recruitment in response to a lung inflation. In five sheep, lung injury was induced by lavage with 0.2% polysorbate 80 in saline. Positron emission tomography and [13N]nitrogen were used to assess regional lung function in dependent, middle, and nondependent lung regions. Physiologic data and positron emission scans were collected before and 5 min after a sustained inflation (continuous positive airway pressure of 50 cm H2O for 30 s)., Results: All animals showed greater loss of aeration and higher perfusion and shunting blood flow in the dependent region. After the RM, Pao2 decreased in all animals by 35 +/- 22 mmHg (P < 0.05). This decrease in Pao2 was associated with redistribution of pulmonary blood flow from the middle, more aerated region to the dependent, less aerated region (P < 0.05) and with an increase in the fraction of pulmonary blood flow that was shunted in the dependent region (P < 0.05). Neither respiratory compliance nor aeration of the dependent region improved after the RM., Conclusions: When a sustained inflation does not restore aeration to atelectatic regions, it can worsen oxygenation by increasing the fraction of pulmonary blood flow that is shunted in nonaerated regions.

A mathematical model was developed to estimate right-to-left shunt (Fs) and the volume of distribution of 13NN in alveolar gas (VA) and shunt tissue (Vs). The data obtained from this model are complementary to, and obtained simultaneously with, pulmonary functional positron emission tomography (PET). The model describes 13NN kinetics in four compartments: central mixing volume, gas-exchanging lung, shunting compartment, and systemic recirculation. To validate the model, five normal prone (NP) and six surfactant-depleted sheep in the supine (LS) and prone (LP) positions were studied under general anesthesia. A central venous bolus of 13NN-labeled saline was injected at the onset of apnea as PET imaging and arterial 13NN sampling were initiated. The model fit the tracer kinetics well (mean r2 = 0.93). Monte Carlo simulations showed that parameters could be accurately identified in the presence of expected experimental noise. Fs derived from the model correlated well with shunt estimates derived from O2 blood concentrations and from PET images. Fs was higher for LS (54 +/- 18%) than for LP (5 +/- 4%) and NP (1 +/- 1%, P < 0.01). VA, as a fraction of PET-measured lung gas volume, was lower for LS (0.18 +/- 0.09) than for LP (0.96 +/- 0.28, P < 0.01), whereas Vs, as a fraction of PET-measured lung tissue volume, was higher for LS (0.46 +/- 0.26) than for LP (0.05 +/- 0.08, P < 0.01). The main conclusions are as follows: 1) the model accurately describes measured arterial 13NN kinetics and provides estimates of Fs, and 2) in this animal model of acute lung injury, the fraction of available gas volume participating in gas exchange is reduced in the supine position.

Techniques of tracheal gas insufflation (TGI) have been shown to enhance CO(2) clearance efficiency in mechanically ventilated patients with acute respiratory distress syndrome (ARDS). Clinical studies have explored the effects of such techniques only at moderate intratracheal gas flow rates, with TGI superimposed to mechanical ventilation in a continuous fashion, or synchronized to the expiratory phase of the duty cycle. We examined the effects of intratracheal pulmonary ventilation (ITPV), delivering the entire tidal volume (VT) in the proximity of the tracheal carina, with all the gas flow supplied continuously through a reverse-thrust catheter (RTC). A potential limitation in the application of TGI is dynamic hyperinflation. Therefore, in a subgroup of patients, we also evaluated the effects of ITPV on end-expiratory lung volume (EELV) by respiratory inductive plethysmography (RIP). Eleven patients with ARDS under volume-cycled mechanical ventilation were subsequently switched to ITPV at the same baseline respiratory rate, I:E ratio, and VT. At the same minute volume, Pa(CO(2)) decreased from 70 +/- 12.3 to 59 +/- 9.5 mm Hg, with a percent reduction of 15 +/- 4% (range from 10 to 20%). The CO(2) decrease was greater in patients with higher baseline Pa(CO(2)) levels (DeltaPa(CO(2)) = 0.29 x Pa(CO(2)) - 9.48, r = 0.95). During transition from mechanical ventilation to ITPV, tracheal positive end-expiratory pressure (PEEP(tr)) decreased with a correspondent decrease in EELV. Both were restored by increasing the PEEP at the ventilator by 3.6 +/- 2.0 cm H(2)O. These data suggest that in patients with ARDS ITPV effectively reduces dead space ventilation and the employment of the RTC may limit or avoid dynamic hyperinflation.

Plasma leakage has been regarded as the main technical problem during prolonged extracorporeal circulation (ECC) with microporous membrane oxygenators (MMOs). We report the case of a 15 year old male who underwent long term ECC for ARDS and in whom, by using new MMOs with reduced pore size, we were able to achieve prolonged artificial gas exchange efficiency with minimal plasma leakage. We conclude that reduced pore size MMOs might represent a valuable technical advance in extracorporeal oxygenation therapy.

Study Objective: To study the effect of positive end-expiratory pressure (PEEP) on the decay of respiratory system compliance (Cpl,rs) due to low tidal volume (VT) ventilation in acute lung injury (ALI) patients., Setting: General ICU in a university hospital., Participants: Eight ALI patients with a lung injury score greater than 2.5., Intervention: Pressure-controlled ventilation (PCV) and volume-controlled ventilation (VCV), with an average VT of 8.5 +/- 0.4 mL/kg, were applied at three levels of PEEP (5, 10, and 15 cm H2O). Before each PCV and VCV period, lung volume history was standardized by manual hyperinflation maneuvers., Measurements: We measured Cpl,rs at time 0 (start), 10, 20, and 30 (end) min from the beginning of each PCV and VCV period. Gas exchange and hemodynamic data were collected at end., Results: At PEEP 5 and 10 cm H2O, we observed a progressive Cpl,rs decay with both PCV and VCV modes. At PEEP 5 cm H2O, we detected a higher Cpl,rs decrease during PCV, due to a higher Cpl,rs at start, compared with VCV. At PEEP 15 cm H2O, Cpl,rs did not decrease significantly. Cpl,rs values measured at end as well as oxygenation and hemodynamic data did not differ between PCV and VCV. At PEEP 15 cm H2O, PCV provided lower PaCO2 than VCV., Conclusions: A PEEP of at least 15 cm H2O was needed to prevent Cpl,rs decay. The progressive Cpl,rs loss we observed at lower PEEP probably reflects alveolar instability.