Selected article for: "accuracy sensitivity specificity and logistic regression model"
Author: Giannakis, A.; Móré, D.; Erdmann, S.; Kintzelé, L.; Fischer, R.M.; Vogel, M.N.; Mangold, D.L.; von Stackelberg, O.; Schnitzler, P.; Zimmermann, S.; Heussel, C.P.; Kauczor, H.U.; Hellbach, K.
Title: COVID-19 pneumonia and its lookalikes: How radiologists perform in differentiating atypical pneumonias
  • Cord-id: oottxkxy
  • Document date: 2021_10_19
    • ID: oottxkxy
    Snippet: PURPOSE: To examine the performance of radiologists in differentiating COVID-19 from non-COVID-19 atypical pneumonia and to perform an analysis of CT patterns in a study cohort including viral, fungal and atypical bacterial pathogens. METHODS: Patients with positive RT-PCR tests for COVID-19 pneumonia (n = 90) and non-COVID-19 atypical pneumonia (n = 294) were retrospectively included. Five radiologists, blinded to the pathogen test results, assessed the CT scans and classified them as COVID-19
    Document: PURPOSE: To examine the performance of radiologists in differentiating COVID-19 from non-COVID-19 atypical pneumonia and to perform an analysis of CT patterns in a study cohort including viral, fungal and atypical bacterial pathogens. METHODS: Patients with positive RT-PCR tests for COVID-19 pneumonia (n = 90) and non-COVID-19 atypical pneumonia (n = 294) were retrospectively included. Five radiologists, blinded to the pathogen test results, assessed the CT scans and classified them as COVID-19 or non-COVID-19 pneumonia. For both groups specific CT features were recorded and a multivariate logistic regression model was used to calculate their ability to predict COVID-19 pneumonia. RESULTS: The radiologists differentiated between COVID-19 and non-COVID-19 pneumonia with an overall accuracy, sensitivity, and specificity of 88% ± 4 (SD), 79% ± 6 (SD), and 90% ± 6 (SD), respectively. The percentage of correct ratings was lower in the early and late stage of COVID-19 pneumonia compared to the progressive and peak stage (68 and 71% vs 85 and 89%). The variables associated with the most increased risk of COVID-19 pneumonia were band like subpleural opacities (OR 5.55, p < 0.001), vascular enlargement (OR 2.63, p = 0.071), and subpleural curvilinear lines (OR 2.52, p = 0.021). Bronchial wall thickening and centrilobular nodules were associated with decreased risk of COVID-19 pneumonia with OR of 0.30 (p = 0.013) and 0.10 (p < 0.001), respectively. CONCLUSIONS: Radiologists can differentiate between COVID-19 and non-COVID-19 atypical pneumonias at chest CT with high overall accuracy, although a lower performance was observed in the early and late stage of COVID 19 pneumonia. Specific CT features might help to make the correct diagnosis.

    Search related documents:
    Co phrase search for related documents
    • accuracy specificity and acute pulmonary disease: 1, 2
    • accuracy specificity and acute setting: 1, 2
    • accuracy specificity and additional information: 1
    • accuracy specificity and logistic regression model: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
    • accuracy specificity sensitivity and acute pulmonary disease: 1
    • accuracy specificity sensitivity and acute setting: 1, 2
    • accuracy specificity sensitivity and logistic regression model: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15
    • acute pneumonia and lobar atelectasis: 1
    • acute pneumonia and logistic regression model: 1, 2, 3, 4, 5, 6
    • acute pulmonary disease and logistic regression model: 1
    • acute setting and additional information: 1, 2
    • acute setting and logistic regression model: 1, 2, 3, 4, 5
    • additional information and logistic regression model: 1, 2, 3