No abstract

Objectives Identifying early markers of poor prognosis of coronavirus disease 2019 (COVID-19) is mandatory. Our purpose is to analyze by chest radiography if rapid worsening of COVID-19 pneumonia in the initial days has predictive value for ventilatory support (VS) need. Methods Ambispective observational ethically approved study in COVID-19 pneumonia inpatients, validated in a second outpatient sample. Brixia score (BS) was applied to the first and second chest radiography required for suspected COVID-19 pneumonia to determine the predictive capacity of BS worsening for VS need. Intraclass correlation coefficient (ICC) was previously analyzed among three radiologists. Sensitivity, specificity, likelihood ratios, AUC, and odds ratio were calculated using ROC curves and binary logistic regression analysis. A value of p < .05 was considered statistically significant. Results A total of 120 inpatients (55 ± 14 years, 68 men) and 112 outpatients (56 ± 13 years, 61 men) were recruited. The average ICC of the BS was between 0.812 (95% confidence interval 0.745–0.878) and 0.906 (95% confidence interval 0.844–0.940). According to the multivariate analysis, a BS worsening per day > 1.3 points within 10 days of the onset of symptoms doubles the risk for requiring VS in inpatients and 5 times in outpatients ( p < .001). The findings from the second chest radiography were always better predictors of VS requirement than those from the first one. Conclusion The early radiological worsening of SARS-CoV-2 pneumonia after symptoms onset is a determining factor of the final prognosis. In elderly patients with some comorbidity and pneumonia, a 48–72-h follow-up radiograph is recommended. Key Points • An early worsening on chest X-ray in patients with SARS-CoV-2 pneumonia is highly predictive of the need for ventilatory support. • This radiological worsening rate can be easily assessed by comparing the first and the second chest X-ray. • In elderly patients with some comorbidity and SARS-CoV-2 pneumonia, close early radiological follow-up is recommended. Supplementary Information The online version contains supplementary material available at 10.1007/s00330-021-08418-3.

No abstract

No abstract

Background The role of chest radiography in COVID-19 disease has changed since the beginning of the pandemic from a diagnostic tool when microbiological resources were scarce to a different one focused on detecting and monitoring COVID-19 lung involvement. Using chest radiographs, early detection of the disease is still helpful in resource-poor environments. However, the sensitivity of a chest radiograph for diagnosing COVID-19 is modest, even for expert radiologists. In this paper, the performance of a deep learning algorithm on the first clinical encounter is evaluated and compared with a group of radiologists with different years of experience. Methods The algorithm uses an ensemble of four deep convolutional networks, Ensemble4Covid, trained to detect COVID-19 on frontal chest radiographs. The algorithm was tested using images from the first clinical encounter of positive and negative cases. Its performance was compared with five radiologists on a smaller test subset of patients. The algorithm's performance was also validated using the public dataset COVIDx. Results Compared to the consensus of five radiologists, the Ensemble4Covid model achieved an AUC of 0.85, whereas the radiologists achieved an AUC of 0.71. Compared with other state-of-the-art models, the performance of a single model of our ensemble achieved nonsignificant differences in the public dataset COVIDx. Conclusion The results show that the use of images from the first clinical encounter significantly drops the detection performance of COVID-19. The performance of our Ensemble4Covid under these challenging conditions is considerably higher compared to a consensus of five radiologists. Artificial intelligence can be used for the fast diagnosis of COVID-19.