RESUMEN
RATIONALE AND OBJECTIVES: Few reports have studied lung aeration and perfusion in normal lungs, COVID-19, and ARDS from other causes (NC-ARDS) using dual-energy computed tomography pulmonary angiograms (DE-CTPA). To describe lung aeration and blood-volume distribution using DE-CTPAs of patients with NC-ARDS, COVID-19, and controls with a normal DE-CTPA ("healthy lungs"). We hypothesized that each of these conditions has unique ranges of aeration and pulmonary blood volumes. MATERIALS AND METHODS: This retrospective, single-center study of DE-CTPAs included patients with COVID-19, NC-ARDS (Berlin criteria), and controls. Patients with macroscopic pulmonary embolisms were excluded. The outcomes studied were the (1) lung blood-volume in areas with different aeration levels (normal, ground glass opacities [GGO], consolidated lung) and (2) aeration/blood-volume ratios. RESULTS: Included were 20 patients with COVID-19 (10 milds, 10 moderate-severe), six with NC-ARDS, and 12 healthy-controls. Lung aeration was lowest in patients with severe COVID-19 24% (IQR13%-31%) followed by those with NC-ARDS 40%(IQR21%-46%). Blood-volume in GGO was lowest in patients with COVID-19 [moderate-severe:-28.6 (IQR-33.1-23.2); mild: -30.1 (IQR-33.3-23.4)] and highest in normally aerated areas in NC-ARDS -37.4 (IQR-52.5-30.2-) and moderate-severe COVID-19 -33.5(IQR-44.2-28.5). The median aeration/blood-volume ratio was lowest in severe COVID-19 but some values overlapped with those observed among patients with NC-ARDS. CONCLUSION: Severe COVID-19 disease is associated with low total aerated lung volume and blood-volume in areas with GGO and overall aeration/blood volume ratios, and with high blood volume in normal lung areas. In this hypothesis-generating study, these findings were most pronounced in severe COVID disease. Larger studies are needed to confirm these preliminary findings.
RESUMEN
OBJECTIVES: In the midst of the coronavirus disease 2019 (COVID-19) outbreak, chest X-ray (CXR) imaging is playing an important role in diagnosis and monitoring of patients with COVID-19. We propose a deep learning model for detection of COVID-19 from CXRs, as well as a tool for retrieving similar patients according to the model's results on their CXRs. For training and evaluating our model, we collected CXRs from inpatients hospitalized in four different hospitals. METHODS: In this retrospective study, 1384 frontal CXRs, of COVID-19 confirmed patients imaged between March and August 2020, and 1024 matching CXRs of non-COVID patients imaged before the pandemic, were collected and used to build a deep learning classifier for detecting patients positive for COVID-19. The classifier consists of an ensemble of pre-trained deep neural networks (DNNS), specifically, ReNet34, ReNet50¸ ReNet152, and vgg16, and is enhanced by data augmentation and lung segmentation. We further implemented a nearest-neighbors algorithm that uses DNN-based image embeddings to retrieve the images most similar to a given image. RESULTS: Our model achieved accuracy of 90.3%, (95% CI: 86.3-93.7%) specificity of 90% (95% CI: 84.3-94%), and sensitivity of 90.5% (95% CI: 85-94%) on a test dataset comprising 15% (350/2326) of the original images. The AUC of the ROC curve is 0.96 (95% CI: 0.93-0.97). CONCLUSION: We provide deep learning models, trained and evaluated on CXRs that can assist medical efforts and reduce medical staff workload in handling COVID-19. KEY POINTS: ⢠A machine learning model was able to detect chest X-ray (CXR) images of patients tested positive for COVID-19 with accuracy and detection rate above 90%. ⢠A tool was created for finding existing CXR images with imaging characteristics most similar to a given CXR, according to the model's image embeddings.