Acute Respiratory Distress Syndrome (ARDS) is a life-threatening condition in critically ill patients, characterized by severe acute hypoxemia and lung injuries. The complexity of ARDS and its high mortality rates warrant innovative approaches for accurate assessment and early intervention. In this scientific study, we present an advanced methodology to evaluate proximal airway volume (PAV) in a porcine model using optical coherence tomography (OCT) and deep learning techniques. We developed an OCT system capable of capturing changes in mucosa thickness (MT) and proximal airway volume in response to smoke inhalation injury in the porcine model. OCT images were acquired at various time points, including baseline, post-injury, and 24, 48, and 72 hours after injury. A comprehensive dataset was compiled for training and validating the deep learning models. The deep learning approach, employing U-Net, DeepLabv3, and SegNet architectures, demonstrated remarkable efficiency in automated PAV calculation when compared to manual segmentation. The Intersection over Union and Dice similarity coefficient metrics validated the accuracy of the models in delineating proximal airway structures.
Acute Respiratory Distress Syndrome (ARDS) is a heterogenic clinical condition that affects critically-ill patients and is associated with high mortality rates and treatment costs. It is characterized by severe acute hypoxemia and alveolar lung injuries. We previously designed an optical coherence tomography (OCT) system to evaluate the changes in mucosa thickness (MT) and proximal airway volume in a swine model after a smoke inhalation injury. However, the analysis relied on manual segmentation of OCT images. Since the manual segmentation of large amounts of OCT data is time-consuming, tedious, and prone to error, this study aims to assess proximal airway volume (PAV) using an automated method based on deep learning. We use convolutional neural networks (CNN) to calculate PAV in a swine model affected by ARDS. We compare the PAV of the swine affected by ARDS with non-ARDS swine. We evaluate OCT images obtained at baseline (BL), post-injury (PI), 24 hours, 48 hours, and 72 hours after smoke inhalation injury. The neural network is modeled utilizing the U-net architecture. The accuracy is evaluated by computing the Sørensen-Dice similarity coefficient. We also demonstrate the correlation between PAV and MT, PFR values obtained from our previous study.
Acute respiratory distress syndrome (ARDS) is a form of lung injury that is associated with inflammation and increased permeability in the lung. It is characterized by acute arterial hypoxemia. The accurate assessment of the airway damage due to smoke inhalation injury (SII) plays a vital role in facilitating appropriate treatment strategies and improved clinical outcomes. This study evaluates the efficiency and accuracy of a trained neural network in segmenting the pig airway images which is used in the assessment of ARDS caused by smoke inhalation injury (SII). The neural network is modeled after the U-net convolutional neural network and the segmentation accuracy is calculated.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.