RESUMEN

The advanced learning paradigm, learning using privileged information (LUPI), leverages information in training that is not present at the time of prediction. In this study, we developed privileged logistic regression (PLR) models under the LUPI paradigm to detect acute respiratory distress syndrome (ARDS), with mechanical ventilation variables or chest x-ray image features employed in the privileged domain and electronic health records in the base domain. In model training, the objective of privileged logistic regression was designed to incorporate data from the privileged domain and encourage knowledge transfer across the privileged and base domains. An asymptotic analysis was also performed, yielding sufficient conditions under which the addition of privileged information increases the rate of convergence in the proposed model. Results for ARDS detection show that PLR models achieve better classification performances than logistic regression models trained solely on the base domain, even when privileged information is partially available. Furthermore, PLR models demonstrate performance on par with or superior to state-of-the-art models under the LUPI paradigm. As the proposed models are effective, easy to interpret, and highly explainable, they are ideal for other clinical applications where privileged information is at least partially available.

Asunto(s)

RESUMEN

BACKGROUND: SARS-CoV-2 pneumonia can cause significant long-term radiological changes, even resembling pulmonary fibrosis. However, the risk factors for these long-term effects are unknown. This study aims to assess radiological abnormalities and their possible risk factors six months after hospital discharge due to COVID-19 pneumonia. MATERIAL AND METHODS: This cross-sectional study in a tertiary hospital included adults admitted for COVID-19 pneumonia from March 2020 to February 2021, who underwent high-resolution computed tomography (HRCT) scans of the chest six months after hospital discharge. The primary outcome was radiological abnormalities on HRCT, while the main explanatory variables were drawn from the patient's medical history along with the disease course, analytical indicators, and the treatment received during admission. RESULTS: The 189 included patients had a mean age of 61.5 years; 70.9% were male, and hypertension was the main comorbidity (45%). About two-thirds (67.2%) presented acute respiratory distress syndrome (ARDS). Most (97.9%) received systemic corticosteroid therapy, and 81% presented pathological findings on HRCT, most commonly ground glass (63.5%), followed by bronchial dilatation (36%) and subpleural bands (25.4%). The multivariable analysis showed that age was the main risk factor, associated with most radiological changes. Other factors were the duration of corticosteroid therapy for ground glass (adjusted odds ratio [aOR] 1.020) as well as a longer stay in the intensive care unit (ICU) (aOR 1.290) and high levels of IL-6 for bronchial dilation (aOR 1.002). CONCLUSION: Radiological involvement of the lungs six months after COVID-19 pneumonia is frequent, especially ground glass. Elderly patients with prolonged ICU admission and a significant inflammatory response measured by IL-6 are more likely to present worse radiological evolution and are candidates for radiological follow-up after COVID-19 pneumonia.

Asunto(s)

RESUMEN

BACKGROUND: Lung ultrasound (LUS) in an emerging technique used in the intensive care unit (ICU). The derivative LUS aeration score has been shown to have associations with mortality in invasively ventilated patients. This study assessed the predictive value of baseline and early changes in LUS aeration scores in critically ill invasively ventilated patients with and without ARDS (Acute Respiratory Distress Syndrome) on 30- and 90-day mortality. METHODS: This is a post hoc analysis of a multicenter prospective observational cohort study, which included patients admitted to the ICU with an expected duration of ventilation for at least 24 h. We restricted participation to patients who underwent a 12-region LUS exam at baseline and had the primary endpoint (30-day mortality) available. Logistic regression was used to analyze the primary and secondary endpoints. The analysis was performed for the complete patient cohort and for predefined subgroups (ARDS and no ARDS). RESULTS: A total of 442 patients were included, of whom 245 had a second LUS exam. The baseline LUS aeration score was not associated with mortality (1.02 (95% CI: 0.99 - 1.06), p = 0.143). This finding was not different in patients with and in patients without ARDS. Early deterioration of the LUS score was associated with mortality (2.09 (95% CI: 1.01 - 4.3), p = 0.046) in patients without ARDS, but not in patients with ARDS or in the complete patient cohort. CONCLUSION: In this cohort of critically ill invasively ventilated patients, the baseline LUS aeration score was not associated with 30- and 90-day mortality. An early change in the LUS aeration score was associated with mortality, but only in patients without ARDS. TRIAL REGISTRATION: ClinicalTrials.gov, ID NCT04482621.

Asunto(s)

RESUMEN

Background: Many survivors of severe COVID-19 pneumonia experience lingering respiratory issues. There is limited research on follow-up chest imaging findings in patients with COVID-19 ARDS, particularly in relation to their mMRC dyspnea scores and pulmonary function tests (PFTs). This study addresses this gap by investigating the clinical characteristics, mMRC dyspnea scores, PFTs, and chest CT findings of COVID-19 ARDS patients at the 6 months post-recovery. By analyzing these variables together, we aim to gain a better understanding of the long-term health consequences of COVID-19 ARDS. Methods: This prospective observational study included 56 subjects with COVID-19 ARDS with dyspnea at the six-month follow-up visits. These patients were evaluated by chest CT, mMRC dyspnea scale, and PFT. The CT severity score was calculated individually for each of the four major imaging findings - ground glass opacities (GGOs), parenchymal/atelectatic bands, reticulations/septal thickening, and consolidation - using a modified CT severity scoring system. Statistics were carried out to find any association between individual CT chest findings and the mMRC dyspnea scale and forced vital capacity (FVC). p values < 0.05 were considered statistically significant. Results: Our study population had a mean age of 55.86 ± 9.60 years, with 44 (78.6%) being men. Grades 1, 2, 3, and 4 on the mMRC dyspnea scale were seen in 57.1%, 30.4%, 10.7%, and 1.8% of patients respectively. Common CT findings observed were GGOs (94.6%), reticulations/septal thickening (96.4%), parenchymal/atelectatic bands (92.8%), and consolidation (14.3%). The mean modified CT severity scores for GGOs, reticulations/septal thickening, parenchymal/atelectatic bands, and consolidation were 10.32 ± 5.51 (range: 0-21), 7.66 ± 4.33 (range: 0-19), 4.77 ± 3.03 (range: 0-14) and 0.29 ± 0.91 (range 0-5) respectively. Reticulations/septal thickening (p = 0.0129) and parenchymal/atelectatic bands (p = 0.0453) were associated with an increased mMRC dyspnea scale. Parenchymal/atelectatic bands were also associated with abnormal FVC (<80%) (p = 0.0233). Conclusion: Six-month follow-up chest CTs of COVID-19 ARDS survivors with persistent respiratory problems showed a statistically significant relationship between increased mMRC dyspnea score and imaging patterns of reticulations/septal thickening and parenchymal/atelectatic bands; while parenchymal/atelectatic bands also showed a statistically significant correlation with reduced FVC.

Asunto(s)

Asunto(s)

RESUMEN

Pulmonary contusion (PC), defined as damage to the lung parenchyma with edema and hemorrhage, has classically been associated with acceleration-deceleration injuries. It is a frequent pathology in clinical practice. However, its clinical presentation and imaging findings are nonspecific. Patients with this entity can present with findings that can range from mild dyspnea to life-threatening respiratory failure and hemodynamic instability. We present the case of a 61-year-old man, a former smoker, who presented to the emergency department after suffering blunt chest trauma. On admission, he complained of only mild shortness of breath, and his vital signs were typical. Initial imaging identified asymmetric pulmonary infiltrates and mediastinal lymphadenopathy; this was suspicious for additional pathology in addition to PC. After an exhaustive evaluation, a neoplastic or infectious disease process was ruled out. Even though the patient presented with a clinical deterioration of respiratory function compatible with secondary acute respiratory distress syndrome, there was a complete recovery after supportive measures and supplemental oxygen. In conclusion, the nonspecific clinical and imaging findings in patients with pulmonary contusion warrant a complete evaluation of these cases. An early diagnosis is essential to establish adequate support and monitoring to prevent possible complications that could worsen the patient's prognosis.

Asunto(s)

RESUMEN

BACKGROUND: Acute respiratory distress syndrome (ARDS) is a life-threatening respiratory condition with high mortality rates, accounting for 10% of all intensive care unit admissions. Lung ultrasound (LUS) as diagnostic tool for acute respiratory failure has garnered widespread recognition and was recently incorporated into the updated definitions of ARDS. This raised the hypothesis that LUS is a reliable method for diagnosing ARDS. OBJECTIVES: We aimed to establish the accuracy of LUS for ARDS diagnosis and classification of focal versus non-focal ARDS subphenotypes. METHODS: This systematic review and meta-analysis used a systematic search strategy, which was applied to PubMed, EMBASE and cochrane databases. Studies investigating the diagnostic accuracy of LUS compared to thoracic CT or chest radiography (CXR) in ARDS diagnosis or focal versus non-focal subphenotypes in adult patients were included. Quality of studies was evaluated using the QUADAS-2 tool. Statistical analyses were performed using "Mada" in Rstudio, version 4.0.3. Sensitivity and specificity with 95% confidence interval of each separate study were summarized in a Forest plot. RESULTS: The search resulted in 2648 unique records. After selection, 11 reports were included, involving 2075 patients and 598 ARDS cases (29%). Nine studies reported on ARDS diagnosis and two reported on focal versus non-focal ARDS subphenotypes classification. Meta-analysis showed a pooled sensitivity of 0.631 (95% CI 0.450-0.782) and pooled specificity of 0.942 (95% CI 0.856-0.978) of LUS for ARDS diagnosis. In two studies, LUS could accurately differentiate between focal versus non-focal ARDS subphenotypes. Insufficient data was available to perform a meta-analysis. CONCLUSION: This review confirms the hypothesis that LUS is a reliable method for diagnosing ARDS in adult patients. For the classification of focal or non-focal subphenotypes, LUS showed promising results, but more research is needed.

Asunto(s)

RESUMEN

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.

Asunto(s)

RESUMEN

Normally aerated lung tissue on computed tomography (CT) is correlated with static respiratory system compliance (Crs) at zero end-expiratory pressure. In clinical practice, however, patients with acute respiratory failure are often managed using elevated PEEP levels. No study has validated the relationship between lung volume and tissue and Crs at the applied positive end-expiratory pressure (PEEP). Therefore, this study aimed to demonstrate the relationship between lung volume and tissue on CT and Crs during the application of PEEP for the clinical management of patients with acute respiratory distress syndrome due to COVID-19. Additionally, as a secondary outcome, the study aimed to evaluate the relationship between CT characteristics and Crs, considering recruitability using the recruitment-to-inflation ratio (R/I ratio). We analyzed the CT and respiratory mechanics data of 30 patients with COVID-19 who were mechanically ventilated. The CT images were acquired during mechanical ventilation at PEEP level of 15 cmH2O and were quantitatively analyzed using Synapse Vincent system version 6.4 (Fujifilm Corporation, Tokyo, Japan). Recruitability was stratified into two groups, high and low recruitability, based on the median R/I ratio of our study population. Thirty patients were included in the analysis with the median R/I ratio of 0.71. A significant correlation was observed between Crs at the applied PEEP (median 15 [interquartile range (IQR) 12.2, 15.8]) and the normally aerated lung volume (r = 0.70 [95% CI 0.46-0.85], P < 0.001) and tissue (r = 0.70 [95% CI 0.46-0.85], P < 0.001). Multivariable linear regression revealed that recruitability (Coefficient = - 390.9 [95% CI - 725.0 to - 56.8], P = 0.024) and Crs (Coefficient = 48.9 [95% CI 32.6-65.2], P < 0.001) were significantly associated with normally aerated lung volume (R-squared: 0.58). In this study, Crs at the applied PEEP was significantly correlated with normally aerated lung volume and tissue on CT. Moreover, recruitability indicated by the R/I ratio and Crs were significantly associated with the normally aerated lung volume. This research underscores the significance of Crs at the applied PEEP as a bedside-measurable parameter and sheds new light on the link between recruitability and normally aerated lung.

Asunto(s)

RESUMEN

BACKGROUND: End-expiratory lung volume (EELV) has been observed to decrease in acute respiratory distress syndrome (ARDS). Yet, research investigating EELV in patients with COVID-19 associated ARDS (CARDS) remains limited. It is unclear whether EELV could serve as a potential metric for monitoring disease progression and identifying patients with ARDS at increased risk of adverse outcomes. STUDY DESIGN AND METHODS: This retrospective study included mechanically ventilated patients diagnosed with CARDS during the initial phase of epidemic control in Shanghai. EELV was measured using the nitrogen washout-washin technique within 48 h post-intubation, followed by regular assessments every 3-4 days. Chest CT scans, performed within a 24-hour window around each EELV measurement, were analyzed using AI software. Differences in patient demographics, clinical data, respiratory mechanics, EELV, and chest CT findings were assessed using linear mixed models (LMM). RESULTS: Out of the 38 patients enrolled, 26.3% survived until discharge from the ICU. In the survivor group, EELV, EELV/predicted body weight (EELV/PBW) and EELV/predicted functional residual capacity (EELV/preFRC) were significantly higher than those in the non-survivor group (survivor group vs. non-survivor group: EELV: 1455 vs. 1162 ml, P = 0.049; EELV/PBW: 24.1 vs. 18.5 ml/kg, P = 0.011; EELV/preFRC: 0.45 vs. 0.34, P = 0.005). Follow-up assessments showed a sustained elevation of EELV/PBW and EELV/preFRC among the survivors. Additionally, EELV exhibited a positive correlation with total lung volume and residual lung volume, while demonstrating a negative correlation with lesion volume determined through chest CT scans analyzed using AI software. CONCLUSION: EELV is a useful indicator for assessing disease severity and monitoring the prognosis of patients with CARDS.

Asunto(s)

RESUMEN

BACKGROUND: ARDS is a heterogeneous syndrome with distinct clinical phenotypes. Here we investigate whether the presence or absence of large pulmonary ultrasonographic consolidations can categorize COVID-19 ARDS patients requiring mechanical ventilation into distinct clinical phenotypes. METHODS: This is a retrospective study performed in a tertiary-level intensive care unit in Israel between April and September 2020. Data collected included lung ultrasound (LUS) findings, respiratory parameters, and treatment interventions. The primary outcome was a composite of three ARDS interventions: prone positioning, high PEEP, or a high dose of inhaled nitric oxide. RESULTS: A total of 128 LUS scans were conducted among 23 patients. The mean age was 65 and about two-thirds were males. 81 scans identified large consolidation and were classified as "C-type", and 47 scans showed multiple B-lines with no or small consolidation and were classified as "B-type". The presence of a "C-type" study had 2.5 times increased chance of receiving the composite primary outcome of advanced ARDS interventions despite similar SOFA scores, Pao2/FiO2 ratio, and markers of disease severity (OR = 2.49, %95CI 1.40-4.44). CONCLUSION: The presence of a "C-type" profile with LUS consolidation potentially represents a distinct COVID-19 ARDS subphenotype that is more likely to require aggressive ARDS interventions. Further studies are required to validate this phenotype in a larger cohort and determine causality, diagnostic, and treatment responses.

Asunto(s)

RESUMEN

BACKGROUND: Acute respiratory distress syndrome (ARDS) is a frequent cause of hypoxemic respiratory failure with a mortality rate of approximately 30%. Identifying ARDS subphenotypes based on "focal" or "non-focal" lung morphology has the potential to better target mechanical ventilation strategies of individual patients. However, classifying morphology through chest radiography or computed tomography is either inaccurate or impractical. Lung ultrasound (LUS) is a non-invasive bedside tool that can accurately distinguish "focal" from "non-focal" lung morphology. We hypothesize that LUS-guided personalized mechanical ventilation in ARDS patients leads to a reduction in 90-day mortality compared to conventional mechanical ventilation. METHODS: The Personalized Mechanical Ventilation Guided by UltraSound in Patients with Acute Respiratory Distress Syndrome (PEGASUS) study is an investigator-initiated, international, randomized clinical trial (RCT) that plans to enroll 538 invasively ventilated adult intensive care unit (ICU) patients with moderate to severe ARDS. Eligible patients will receive a LUS exam to classify lung morphology as "focal" or "non-focal". Thereafter, patients will be randomized within 12 h after ARDS diagnosis to receive standard care or personalized ventilation where the ventilation strategy is adjusted to the morphology subphenotype, i.e., higher positive end-expiratory pressure (PEEP) and recruitment maneuvers for "non-focal" ARDS and lower PEEP and prone positioning for "focal" ARDS. The primary endpoint is all-cause mortality at day 90. Secondary outcomes are mortality at day 28, ventilator-free days at day 28, ICU length of stay, ICU mortality, hospital length of stay, hospital mortality, and number of complications (ventilator-associated pneumonia, pneumothorax, and need for rescue therapy). After a pilot phase of 80 patients, the correct interpretation of LUS images and correct application of the intervention within the safe limits of mechanical ventilation will be evaluated. DISCUSSION: PEGASUS is the first RCT that compares LUS-guided personalized mechanical ventilation with conventional ventilation in invasively ventilated patients with moderate and severe ARDS. If this study demonstrates that personalized ventilation guided by LUS can improve the outcomes of ARDS patients, it has the potential to shift the existing one-size-fits-all ventilation strategy towards a more individualized approach. TRIAL REGISTRATION: The PEGASUS trial was registered before the inclusion of the first patient, https://clinicaltrials.gov/ (ID: NCT05492344).

Asunto(s)

Asunto(s)

RESUMEN

Acute respiratory distress syndrome (ARDS) is characterized by a redistribution of regional lung perfusion that impairs gas exchange. While speculative, experimental evidence suggests that perfusion redistribution may contribute to regional inflammation and modify disease progression. Unfortunately, tools to visualize and quantify lung perfusion in patients with ARDS are lacking. This review explores recent advances in perfusion imaging techniques that aim to understand the pulmonary circulation in ARDS. Dynamic contrast-enhanced computed tomography captures first-pass kinetics of intravenously injected dye during continuous scan acquisitions. Different contrast characteristics and kinetic modeling have improved its topographic measurement of pulmonary perfusion with high spatial and temporal resolution. Dual-energy computed tomography can map the pulmonary blood volume of the whole lung with limited radiation exposure, enabling its application in clinical research. Electrical impedance tomography can obtain serial topographic assessments of perfusion at the bedside in response to treatments such as inhaled nitric oxide and prone position. Ongoing technological improvements and emerging techniques will enhance lung perfusion imaging and aid its incorporation into the care of patients with ARDS.

Asunto(s)

RESUMEN

Acute respiratory distress syndrome (ARDS) continues to cause significant morbidity and mortality worldwide. However, there are currently no corresponding specific medications. Lung, right heart, and diaphragm protective therapy is the cornerstone of ARDS management. Non-invasive, radiation-free, and transportation-free point-of-care ultrasound enables for real-time dynamic evaluation of the lung, right heart, and diaphragm. It is an essential tool for the diagnosis, monitoring, and clinical decision-making of ARDS patients. However, ultrasound has several limitations, and its clinical application must be carefully evaluated in conjunction with clinical symptoms and other monitoring techniques. Further research is needed to determine whether clinical decision-making based on point-of-care ultrasound examination can improve the prognosis of ARDS patients.

Asunto(s)

RESUMEN

Introduction: Systematic evaluation of long-term outcomes in survivors of H1N1 is still lacking. This study aimed to characterize long-term outcomes of severe H1N1-induced pneumonia and acute respiratory distress syndrome (ARDS). Method: This was a single-center, prospective, cohort study. Survivors were followed up for four times after discharge from intensive care unit (ICU) by lung high-resolution computed tomography (HRCT), pulmonary function assessment, 6-minute walk test (6MWT), and SF-36 instrument. Result: A total of 60 survivors of H1N1-induced pneumonia and ARDS were followed up for four times. The carbon monoxide at single breath (DLCO) of predicted values and the 6MWT results didn't continue improving after 3 months. Health-related quality of life didn't change during the 12 months after ICU discharge. Reticulation or interlobular septal thickening on HRCT did not begin to improve significantly until the 12-month follow-up. The DLCO of predicted values showed negative correlation with the severity degree of primary disease and reticulation or interlobular septal thickening, and a positive correlation with physical functioning. The DLCO of predicted values and reticulation or interlobular septal thickening both correlated with the highest tidal volume during mechanical ventilation. Levels of fibrogenic cytokines had a positive correlation with reticulation or interlobular septal thickening. Conclusion: The improvements in pulmonary function and exercise capacity, imaging, and health-related quality of life had different time phase and impact on each other during 12 months of follow-up. Long-term outcomes of pulmonary fibrosis might be related to the lung injury and excessive lung fibroproliferation at the early stage during ICU admission.

Asunto(s)

RESUMEN

BACKGROUND: Cerebral microbleeds (CMBs) have been described in critically ill patients with respiratory failure, acute respiratory distress syndrome (ARDS), or sepsis. This scoping review aimed to systematically summarize existing literature on critical illness-associated CMBs. METHODS: Studies reporting on adults admitted to the intensive care unit for respiratory failure, ARDS, or sepsis with evidence of CMBs on magnetic resonance imaging were included for review following a systematic search across five databases (MEDLINE, Embase, Cochrane Central Register of Controlled Trials (CENTRAL), Scopus, and Web of Science) and a two-stage screening process. Studies were excluded if patients' CMBs were clearly explained by another process of neurological injury. RESULTS: Forty-eight studies reporting on 216 critically ill patients (mean age 57.9, 18.4% female) with CMBs were included. Of 216, 197 (91.2%) patients developed respiratory failure or ARDS, five (2.3%) patients developed sepsis, and 14 (6.5%) patients developed both respiratory failure and sepsis. Of 211 patients with respiratory failure, 160 (75.8%) patients had coronavirus disease 2019. The prevalence of CMBs among critically ill patients with respiratory failure or ARDS was 30.0% (111 of 370 patients in cohort studies). The corpus callosum and juxtacortical area were the most frequently involved sites for CMBs (64.8% and 41.7% of all 216 patients, respectively). Functional outcomes were only reported in 48 patients, among whom 31 (64.6%) were independent at discharge, four (8.3%) were dependent at discharge, and 13 (27.1%) did not survive until discharge. Cognitive outcomes were only reported in 11 of 216 patients (5.1%), all of whom showed cognitive deficits (nine patients with executive dysfunction and two patients with memory deficits). CONCLUSIONS: Cerebral microbleeds are commonly reported in patients with critical illness due to respiratory failure, ARDS, or sepsis. CMBs had a predilection for the corpus callosum and juxtacortical area, which may be specific to critical illness-associated CMBs. Functional and cognitive outcomes of these lesions are largely unknown.

Asunto(s)

RESUMEN

OBJECTIVES: Although respiratory failure is the most common feature in coronavirus disease 2019 (COVID-19), abdominal organ involvement is likewise frequently observed. To investigate visceral and thoracic circulation and abdominal organ damage in COVID-19 patients. MATERIALS AND METHODS: A monocentric observational study was carried on. In COVID-19 patients affected by acute respiratory distress syndrome (ARDS) (n = 31) or mild pneumonia (n = 60) thoracoabdominal circulation was evaluated using Doppler-ultrasound and computed tomography. The study also included non-COVID-19 patients affected by ARDS (n = 10) or portal hypertension (n = 10) for comparison of the main circulatory changes. RESULTS: Patients affected by COVID-19 ARDS showed hyperdynamic visceral flow and increased portal velocity, hepatic artery resistance-index, and spleen diameter relative to those with mild-pneumonia (p = 0.001). Splanchnic circulatory parameters significantly correlated with the main respiratory indexes (p < 0.001) and pulmonary artery diameter (p = 0.02). The chest and abdominal vascular remodeling pattern of COVID-19 ARDS patients resembled the picture observed in the PH group, while differed from that of the non-COVID ARDS group. A more severe COVID-19 presentation was associated with worse liver dysfunction and enhanced inflammatory activation; these parameters both correlated with abdominal (p = 0.04) and chest imaging measures (p = 0.03). CONCLUSION: In COVID-19 ARDS patients there are abdominal and lung vascular modifications that depict a portal hypertension-like pattern. The correlation between visceral vascular remodeling, pulmonary artery enlargement, and organ damage in these critically ill patients is consistent with a portal hyperlfow-like syndrome that could contribute to the peculiar characteristics of respiratory failure in these patients. CLINICAL RELEVANCE STATEMENT: our data suggest that the severity of COVID-19 lung involvement is directly related to the development of a portal hyperflow-like syndrome. These observations should help in defining the need for a closer monitoring, but also to develop dedicated therapeutic strategies.

Asunto(s)

RESUMEN

OBJECTIVE: This study aims to design, validate and assess the accuracy a deep learning model capable of differentiation Chest X-Rays between pneumonia, acute respiratory distress syndrome (ARDS) and normal lungs. MATERIALS AND METHODS: A diagnostic performance study was conducted using Chest X-Ray images from adult patients admitted to a medical intensive care unit between January 2003 and November 2014. X-ray images from 15,899 patients were assigned one of three prespecified categories: "ARDS", "Pneumonia", or "Normal". RESULTS: A two-step convolutional neural network (CNN) pipeline was developed and tested to distinguish between the three patterns with sensitivity ranging from 91.8% to 97.8% and specificity ranging from 96.6% to 98.8%. The CNN model was validated with a sensitivity of 96.3% and specificity of 96.6% using a previous dataset of patients with Acute Lung Injury (ALI)/ARDS. DISCUSSION: The results suggest that a deep learning model based on chest x-ray pattern recognition can be a useful tool in distinguishing patients with ARDS from patients with normal lungs, providing faster results than digital surveillance tools based on text reports. CONCLUSION: A CNN-based deep learning model showed clinically significant performance, providing potential for faster ARDS identification. Future research should prospectively evaluate these tools in a clinical setting.

Asunto(s)

Asunto(s)