Does imbalance in chest X-ray datasets produce biased deep learning approaches for COVID-19 screening?

Background

The health crisis resulting from the global COVID-19 pandemic highlighted more than ever the need for rapid, reliable and safe methods of diagnosis and monitoring of respiratory diseases. To study pulmonary involvement in detail, one of the most common resources is the use of different lung imaging modalities (like chest radiography) to explore the possible affected areas.

Methods

The study of patient characteristics like sex and age in pathologies of this type is crucial for gaining knowledge of the disease and for avoiding biases due to the clear scarcity of data when developing representative systems. In this work, we performed an analysis of these factors in chest X-ray images to identify biases. Specifically, 11 imbalance scenarios were defined with female and male COVID-19 patients present in different proportions for the sex analysis, and 6 scenarios where only one specific age range was used for training for the age factor. In each study, 3 different approaches for automatic COVID-19 screening were used: Normal vs COVID-19, Pneumonia vs COVID-19 and Non-COVID-19 vs COVID-19. The study was validated using two public chest X-ray datasets, allowing a reliable analysis to support the clinical decision-making process.

Results

The results for the sex-related analysis indicate this factor slightly affects the system in the Normal VS COVID-19 and Pneumonia VS COVID-19 approaches, although the identified differences are not relevant enough to worsen considerably the system. Regarding the age-related analysis, this factor was observed to be influencing the system in a more consistent way than the sex factor, as it was present in all considered scenarios. However, this worsening does not represent a major factor, as it is not of great magnitude.

Conclusions

Multiple studies have been conducted in other fields in order to determine if certain patient characteristics such as sex or age influenced these deep learning systems. However, to the best of our knowledge, this study has not been done for COVID-19 despite the urgency and lack of COVID-19 chest x-ray images. The presented results evidenced that the proposed methodology and tested approaches allow a robust and reliable analysis to support the clinical decision-making process in this pandemic scenario.

Supplementary Information

The online version contains supplementary material available at (10.1186/s12874-022-01578-w).

Altered Th17/Treg Ratio in Peripheral Blood of Systemic Lupus Erythematosus but Not Primary Antiphospholipid Syndrome

Introduction: The role of the immune response in the pathogenesis of antiphospholipid syndrome (APS) remains elusive. It is possible that differences in the frequencies of Th17 cells and/or defects in the immunoregulatory mechanisms are involved in the pathogenesis of APS. Our aim was to determine the peripheral blood Th cells phenotype and the circulating cytokine profile in patients with primary APS (pAPS) and compare it with systemic lupus erythemathosus (SLE) as disease control group.

Methods: The frequencies of circulating regulatory T cells (Tregs) were determined in PBMCs from 36 patients with pAPS by flow cytometry. As control groups we included 21 age- and gender-matched healthy controls (HC) and 11 patients with SLE. The suppressive capacity of Tregs was evaluated in vitro by coculture assay. On the other hand, intracellular cytokine production was assessed in Th1, Th2, and Th17 cells and circulating IL-6, IL-10, and IL-35 were measured by Cytometric Bead Array and ELISA. The quantification of Th master gene expression levels was performed by real time quantitative PCR.

Results: pAPS patients and SLE patients did not show differences in the percentage or number of Tregs compared to HC. The suppressive capacity of Tregs was also similar in the three study group. Instead, we found higher FoxP3·mRNA expression levels in pAPS patients and HC than SLE patients. Regarding the Th17 response, patients with pAPS and HC showed a significantly lower frequency of circulating Th17 cells than SLE. However, no differences were observed in the Th1 response between patients and controls. Thus, increased Th17/Th1 and Th17/Treg ratios were found in SLE patients but not in pAPS patients. pAPS and SLE patients had higher serum IL-6 levels than HC but there was not difference between both disease groups. Besides, a significant increase in the immunosuppressive cytokine levels was observed only in pAPS as compared to HC.

Conclusions: Our data demonstrate an increased inflammatory profile of peripheral blood CD4⁺ T cells from SLE as compared with pAPS mostly due to an increased Th17 response. In conclusion, there seems not to be a direct pathogenic role for Th cells in pAPS but in SLE.

Phagocyte dysfunction in polymyalgia rheumatica and other age-related, chronic, inflammatory conditions

This study was conducted to evaluate phagocyte function in patients with age-related chronic inflammatory conditions. It included 95 patients with PMR, 17 with GCA, 40 with EORA, and 25 age-matched HCs. Serum IL-8 was determined with a bead array. The chemotactic capacity, phagocytic ability, and oxidative burst activity of circulating leukocytes were determined with flow cytometry kits. Patients with active chronic inflammatory diseases showed a significant increase in circulating levels of IL-8 that remained elevated in patients with PMR or EORA, despite treatment. No correlation was found between circulating IL-8 and the migratory capacity of neutrophils. Neutrophils from patients with active EORA without stimulus and after fMLP stimuli showed a higher capacity to migrate than those of the HCs (P=0.033). The phagocytic activity of granulocytes in the patients with GCA was significantly higher than in the HCs and the patients with PMR or EORA (P<0.05). The percentage and MFI of phagocytes that produce ROIs when stimulated with Escherichia coli was significantly reduced in neutrophils and monocytes from the patients with age-restricted inflammatory conditions. We concluded that the effector functions of phagocytes, determined to be chemotaxis, phagocytosis, and oxidative burst, are deregulated in age-restricted inflammatory disorders and may have a pathogenic role.

Induction of proinflammatory cytokines in human lung epithelial cells during Rhodococcus equi infection

Rhodococcus equi is an opportunistic human pathogen associated with immunosuppressed people. While the interaction of R. equi with macrophages has been comprehensively studied, little is known about its interactions with non-phagocytic cells. Here, we characterized the entry process of this bacterium into human lung epithelial cells. The invasion is inhibited by nocodazole and wortmannin, suggesting that the phosphatidylinositol 3-kinase pathway and microtubule cytoskeleton are important for invasion. Pre-incubation of R. equi with a rabbit anti-R. equi polyclonal antiserum resulted in a dramatic reduction in invasion. Also, the invasion process as studied by immunofluorescence and scanning electron microscopy indicates that R. equi make initial contact with the microvilli of the A549 cells, and at the structural level, the entry process was observed to occur via a zipper-like mechanism. Infected lung epithelial cells upregulate the expression of cytokines IL-8 and IL-6 upon infection. The production of these pro-inflammatory cytokines was significantly enhanced in culture supernatants from cells infected with non-mucoid plasmid-less strains when compared with cells infected with mucoid strains. These results demonstrate that human airway epithelial cells produce pro-inflammatory mediators against R. equi isolates.