COVID-19 classification of X-ray images using deep neural networks.

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.

Vocal cord paresis on CTA - A novel tool for the diagnosis of lateral medullary syndrome.

BACKGROUND: Diagnosis of lateral medullary syndrome (LMS) is often delayed due to elusive clinical presentations and frequently non-revealing neuroimaging tests. We aimed to investigate the use of ipsilateral vocal cord paresis (VCP) identified on neck computed tomography angiography (CTA) as an early diagnostic sign for LMS. METHODS: Medical records were queried for patients admitted with LMS between 1/2012 and 10/2020. A control group of patients undergoing CTA for transient or no neurological symptoms was matched for sex and age. Clinical data were collected by a stroke neurologist. Two neuroradiologists independently and blindly assessed CTA images for radiological signs of VCP. RESULTS: Fifteen LMS and 15 control patients were included in the analysis. Median time from arrival to LMS diagnosis was 29.4 h [IQR 7,47] and twice as long in patients who suffered aspiration pneumonia. Thrombolysis rate was 0% in LMS patients versus 14.5% in general stroke patients. Dysphonia was noted in the emergency department in three (20%) patients, whereas all 15 patients had radiological signs of VCP on CTA. Medialization of a true vocal cord was the most sensitive (100%) and specific (80-87%) sign for LMS, with good inter-rater agreement (kappa 0.66). Timely detection of VCP on CTA could have shortened median time to LMS diagnosis by 14 h and enabled thrombolytic therapy in 3 (20%) patients. CONCLUSIONS: VCP on CTA is a valuable sign for the diagnosis of LMS. If detected early, it may enable reperfusion therapy and prevent aspiration pneumonia, consequently saving life and diminishing disability.