MEDPSeg: Hierarchical polymorphic multitask learning for the segmentation of ground-glass opacities, consolidation, and pulmonary structures on computed tomography (preprint)

The COVID-19 pandemic response highlighted the potential of deep learning methods in facilitating the diagnosis, prognosis and understanding of lung diseases through automated segmentation of pulmonary structures and lesions in chest computed tomography (CT). Automated separation of lung lesion into ground-glass opacity (GGO) and consolidation is hindered due to the labor-intensive and subjective nature of this task, resulting in scarce availability of ground truth for supervised learning. To tackle this problem, we propose MEDPSeg. MEDPSeg learns from heterogeneous chest CT targets through hierarchical polymorphic multitask learning (HPML). HPML explores the hierarchical nature of GGO and consolidation, lung lesions, and the lungs, with further benefits achieved through multitasking airway and pulmonary artery segmentation. Over 6000 volumetric CT scans from different partially labeled sources were used for training and testing. Experiments show PML enabling new state-of-the-art performance for GGO and consolidation segmentation tasks. In addition, MEDPSeg simultaneously performs segmentation of the lung parenchyma, airways, pulmonary artery, and lung lesions, all in a single forward prediction, with performance comparable to state-of-the-art methods specialized in each of those targets. Finally, we provide an open-source implementation with a graphical user interface at https://github.com/MICLab-Unicamp/medpseg.

Small Airway Disease as long-term Sequela of COVID-19: Use of Expiratory CT despite Improvement in Pulmonary Function test (preprint)

Background- It is important to understand the spectrum of pulmonary diseases that patients are presenting after recovery from initial SARS-CoV-2 infection. We aim to study small airway disease and changes in Computed Tomography (CT) and pulmonary function tests (PFTs) with time. Methods: This is retrospective observation study including adult patients with confirmed SARS-CoV-2 infection with at-least two CT scans either during acute (defined as < 1 month) or subacute (1-3 months) or chronic (>3months) phase after positive test. Radiological features and follow up PFTs were obtained. Results: 22 patients met the inclusion criteria with mean age 57.6 years (range 36-83). Out of these,18 (81.81%) were hospitalized. Mean duration of diagnosis to CT and PFT was 192.68 (112-385) days and 161.54 (31-259) days respectively. On PFTs, restrictive pulmonary physiology was predominant finding during subacute 56.25% (9/16) and chronic phases 47% (7/15). PFTs improved significantly with time {FEV1((p=0.0361), FVC (p=0.0341), FEF 25%-75% (p=0.0259) and DLCO (p=0.0019)}, but there was persistent air trapping in the expiratory chronic phase CT. There was resolution of ground glass opacity, consolidation, and bronchiectasis however air trapping increased with time in 41.61% (10/21) of subacute CTs compared to 81.25% (13/16) in chronic CTs. Conclusion - Our study shows evidence of airway as well as parenchymal disease as relatively long-term sequel of SARS-CoV-2 infection. It also highlights the natural course and spontaneous recovery of some radiological and pulmonary function test abnormalities over time with evidence of persistent small airway disease (air trapping) on expiratory CT imaging months after infection.

Small Airways Disease is a Post-Acute Sequelae of SARS-CoV-2 Infection (preprint)

Background The sequelae of SARS-CoV-2 infection on pulmonary structure and function remain incompletely characterized. Methods Adults with confirmed COVID-19 who remained symptomatic more than thirty days following diagnosis were enrolled and classified as ambulatory, hospitalized or requiring the intensive care unit (ICU) based on the highest level of care received during acute infection. Symptoms, pulmonary function tests and chest computed tomography (CT) findings were compared across groups and to healthy controls. CT images were quantitatively analyzed using supervised machine-learning to measure regional ground glass opacities (GGO) and image-matching to measure regional air trapping. Comparisons were performed using univariate analyses and multivariate linear regression. Results Of the 100 patients enrolled, 67 were in the ambulatory group. All groups commonly reported cough and dyspnea. Pulmonary function testing revealed restrictive physiology in the hospitalized and ICU groups but was normal in the ambulatory group. Among hospitalized and ICU patients, the mean percent of total lung classified as GGO was 13.2% and 28.7%, respectively, and was higher than in ambulatory patients (3.7%, P<0.001). The mean percentage of total lung affected by air trapping was 25.4%, 34.5% and 27.2% in the ambulatory, hospitalized and ICU groups and 7.3% in healthy controls (P<0.001). Air trapping measured by quantitative CT correlated with the residual volume to total lung capacity ratio (RV/TLC; {rho} =0.6, P<0.001). Conclusions Air trapping is present in patients with post-acute sequelae of COVID-19 and is independent of initial infection severity, suggesting obstruction at the level of the small airways. The long-term consequences are not known.

Genetic and non-genetic factors affecting the expression of COVID-19 relevant genes in the large airway epithelium (preprint)

Particular host and environmental factors influence susceptibility to severe COVID-19. We analyzed RNA-sequencing data from bronchial epithelial brushings - a relevant tissue for SARS-CoV-2 infection - obtained from three cohorts of uninfected individuals, and investigated how non-genetic and genetic factors affect the regulation of host genes implicated in COVID-19. We found that ACE2 expression was higher in relation to active smoking, obesity, and hypertension that are known risk factors of COVID-19 severity, while an association with interferon-related inflammation was driven by the truncated, non-binding ACE2 isoform. We discovered that expression patterns of a suppressed airway immune response to early SARS-CoV-2 infection, compared to other viruses, are similar to patterns associated with obesity, hypertension, and cardiovascular disease, which may thus contribute to a COVID-19-susceptible airway environment. eQTL mapping identified regulatory variants for genes implicated in COVID-19, some of which had pheWAS evidence for their potential role in respiratory infections. These data provide evidence that clinically relevant variation in the expression of COVID-19-related genes is associated with host factors, environmental exposures, and likely host genetic variation.