ObjectivesWe tested artificial intelligence (AI) to support the diagnosis of COVID-19 using chest X-ray (CXR). Diagnostic performance was computed for a system trained on CXRs of Italian subjects from two hospitals in Lombardy, Italy. MethodsWe used for training and internal testing an ensemble of ten convolutional neural networks (CNNs) with mainly bedside CXRs of 250 COVID-19 and 250 non-COVID-19 subjects from two hospitals.We then tested such system on bedside CXRs of an independent group of 110 patients 36 non-COVID-19) from one of the two hospitals. A retrospective reading was performed by two radiologists in the absence of any clinical information, with the aim to differentiate COVID-19 from non-COVID-19 patients. Real-time polymerase chain reaction served as reference standard.Results At 10-fold cross-validation, our AI model classified COVID-19 and non COVID-19 patients with 0.78 sensitivity (95% confidence interval [CI] 0.74-0.81), 0.82 specificity (95% CI 0.78-0.85) and 0.89 area under the curve (AUC) (95% CI 0.86-0.91). For the independent dataset, AI showed 0.80 sensitivity (95% CI 0.72-0.86) (59/74), 0.81 specificity (29/36) (95% CI 0.73-0.87), and 0.81 AUC (95% CI 0.73-0.87). Radiologists' reading obtained 0.63 sensitivity (95% CI 0.52-0.74) and 0.78 specificity (95% CI 0.61-0.90) in one centre and 0.64 sensitivity (95% CI 0.52-0.74) and 0.86 specificity (95% CI 0.71-0.95) in the other.Conclusions This preliminary experience based on ten CNNs trained on a limited training dataset shows an interesting potential of AI for COVID-19 diagnosis. Such tool is in training with new CXRs to further increase its performance.

Background We aimed to train and test a deep learning classifier to support the diagnosis of coronavirus disease 2019 (COVID-19) using chest x-ray (CXR) on a cohort of subjects from two hospitals in Lombardy, Italy. Methods We used for training and validation an ensemble of ten convolutional neural networks (CNNs) with mainly bedside CXRs of 250 COVID-19 and 250 non-COVID-19 subjects from two hospitals (Centres 1 and 2). We then tested such system on bedside CXRs of an independent group of 110 patients (74 COVID-19, 36 non-COVID-19) from one of the two hospitals. A retrospective reading was performed by two radiologists in the absence of any clinical information, with the aim to differentiate COVID-19 from non-COVID-19 patients. Real-time polymerase chain reaction served as the reference standard. Results At 10-fold cross-validation, our deep learning model classified COVID-19 and non-COVID-19 patients with 0.78 sensitivity (95% confidence interval [CI] 0.74–0.81), 0.82 specificity (95% CI 0.78–0.85), and 0.89 area under the curve (AUC) (95% CI 0.86–0.91). For the independent dataset, deep learning showed 0.80 sensitivity (95% CI 0.72–0.86) (59/74), 0.81 specificity (29/36) (95% CI 0.73–0.87), and 0.81 AUC (95% CI 0.73–0.87). Radiologists’ reading obtained 0.63 sensitivity (95% CI 0.52–0.74) and 0.78 specificity (95% CI 0.61–0.90) in Centre 1 and 0.64 sensitivity (95% CI 0.52–0.74) and 0.86 specificity (95% CI 0.71–0.95) in Centre 2. Conclusions This preliminary experience based on ten CNNs trained on a limited training dataset shows an interesting potential of deep learning for COVID-19 diagnosis. Such tool is in training with new CXRs to further increase its performance.

To evaluate the diagnostic accuracy and the imaging features of routine admission chest X-ray in patients suspected for novel Coronavirus 2019 (SARS-CoV-2) infection. Method: We retrospectively evaluated clinical and X-ray features in all patients referred to the emergency department for suspected SARS-CoV-2 infection between March 1st and March 13th. A single radiologist with more than 15 years of experience in chest-imaging evaluated the presence and extent of alveolar opacities, reticulations, and/or pleural effusion. The percentage of lung involvement (range < 25 % to 75-100 %) was also calculated. We stratified patients in groups according to the time interval between symptoms onset and X-ray imaging (≤ 5 and > 5 days) and according to age (≤ 50 and > 50 years old). Results: A total of 518 patients were enrolled. Overall 314 patients had negative and 204 had positive RT-PCR results. Lung lesions in patients with SARS-Cov2 pneumonia primarily manifested as alveolar and interstitial opacities and were mainly bilateral (60.8 %). Lung abnormalities were more frequent and more severe by symptom duration and by increasing age. The sensitivity and specificity of chest X-ray at admission in the overall cohort were 57 % (95 % CI = 47-67) and 89 % (83-94), respectively. Sensitivity was higher for patients with symptom onset > 5 days compared to ≤ 5 days (76 % [62-87] vs 37 % ) and in patients > 50 years old compared to ≤ 50 years (59 % [48-69] vs 47 % [23-72]), at the expense of a slightly lower specificity (68 % [45-86] and 82 % [73-89], respectively). Conclusions: Overall chest X-ray sensitivity for SARS-CoV-2 pneumonia was 57 %. Sensitivity was higher when symptoms had started more than 5 days before, at the expense of lesser specificity, while slightly higher in older patients in comparison to younger ones.

No abstract