A convolutional neural network-based COVID-19 detection method using chest CT images.

Background: High-throughput population screening for the novel coronavirus disease (COVID-19) is critical to controlling disease transmission. Convolutional neural networks (CNNs) are a cutting-edge technology in the field of computer vision and may prove more effective than humans in medical diagnosis based on computed tomography (CT) images. Chest CT images can show pulmonary abnormalities in patients with COVID-19. Methods: In this study, CT image preprocessing are firstly performed using fuzzy c-means (FCM) algorithm to extracted the region of the pulmonary parenchyma. Through multiscale transformation, the preprocessed image is subjected to multi scale transformation and RGB (red, green, blue) space construction. After then, the performances of GoogLeNet and ResNet, as the most advanced CNN architectures, were compared in COVID-19 detection. In addition, transfer learning (TL) was employed to solve overfitting problems caused by limited CT samples. Finally, the performance of the models were evaluated and compared using the accuracy, recall rate, and F1 score. Results: Our results showed that the ResNet-50 method based on TL (ResNet-50-TL) obtained the highest diagnostic accuracy, with a rate of 82.7% and a recall rate of 79.1% for COVID-19. These results showed that applying deep learning technology to COVID-19 screening based on chest CT images is a very promising approach. This study inspired us to work towards developing an automatic diagnostic system that can quickly and accurately screen large numbers of people with COVID-19. Conclusions: We tested a deep learning algorithm to accurately detect COVID-19 and differentiate between healthy control samples, COVID-19 samples, and common pneumonia samples. We found that TL can significantly increase accuracy when the sample size is limited.

Pulmonary fibrosis and its related factors in discharged patients with new corona virus pneumonia: a cohort study.

BACKGROUND: Thousands of Coronavirus Disease 2019 (COVID-19) patients have been discharged from hospitals Persistent follow-up studies are required to evaluate the prevalence of post-COVID-19 fibrosis. METHODS: This study involves 462 laboratory-confirmed patients with COVID-19 who were admitted to Shenzhen Third People's Hospital from January 11, 2020 to April 26, 2020. A total of 457 patients underwent thin-section chest CT scans during the hospitalization or after discharge to identify the pulmonary lesion. A total of 287 patients were followed up from 90 to 150 days after the onset of the disease, and lung function tests were conducted about three months after the onset. The risk factors affecting the persistence of pulmonary fibrosis were identified through regression analysis and the prediction model of the persistence of pulmonary fibrosis was established. RESULTS: Parenchymal bands, irregular interfaces, reticulation and traction bronchiectasis were the most common CT features in all COVID-19 patients. During the 0-30, 31-60, 61-90, 91-120 and > 120 days after onset, 86.87%, 74.40%, 79.56%, 68.12% and 62.03% patients developed with pulmonary fibrosis and 4.53%, 19.61%, 18.02%, 38.30% and 48.98% patients reversed pulmonary fibrosis, respectively. It was observed that Age, BMI, Fever, and Highest PCT were predictive factors for sustaining fibrosis even after 90 days from onset. A predictive model of the persistence with pulmonary fibrosis was developed based-on the Logistic Regression method with an accuracy, PPV, NPV, Sensitivity and Specificity of the model of 76%, 71%, 79%, 67%, and 82%, respectively. More than half of the COVID-19 patients revealed abnormal conditions in lung function after 90 days from onset, and the ratio of abnormal lung function did not differ on a statistically significant level between the fibrotic and non-fibrotic groups. CONCLUSIONS: Persistent pulmonary fibrosis was more likely to develop in patients with older age, higher BMI, severe/critical condition, fever, a longer viral clearance time, pre-existing disease and delayed hospitalization. Fibrosis developed in COVID-19 patients could be reversed in about a third of the patients after 120 days from onset. The pulmonary function of less than half of COVID-19 patients could turn to normal condition after three months from onset. An effective prediction model with an average area under the curve (AUC) of 0.84 was established to predict the persistence of pulmonary fibrosis in COVID-19 patients for early diagnosis.

Thin-section computed tomography findings and longitudinal variations of the residual pulmonary sequelae after discharge in patients with COVID-19: a short-term follow-up study.

OBJECTIVES: This study analyzed and compared CT findings and longitudinal variations after discharge between severe and non-severe coronavirus disease (COVID-19) patients who had residual pulmonary sequelae at pre-discharge. METHODS: A total of 310 patients were included and stratified into severe and non-severe COVID-19 groups. Cross-sectional CT features across different time periods (T0: pre-discharge, T1: 1-4 weeks after discharge, T2: 5-8 weeks after discharge, T3: 9-12 weeks after discharge, T4: > 12 weeks after discharge) were compared, and the longitudinal variations of CT findings were analyzed and compared in both groups. RESULTS: The cumulative absorption rate of fibrosis-like findings in the severe and non-severe groups at T4 was 24.3% (17/70) and 52.0% (53/102), respectively. In both groups, ground-glass opacity (GGO) with consolidation showed a clear decreasing trend at T1, after which they maintained similar lower levels. The GGO in the severe group showed an increasing trend first at T1 and then decreasing at T4; however, the incidence decreased gradually in the non-severe group. Most fibrosis-like findings showed a tendency to decrease rapidly and then remained stable. Bronchial dilatation in the severe group persisted at an intermediate level. CONCLUSIONS: After discharge, the characteristics and changing trends of pulmonary sequelae caused by COVID-19 were significantly different between the two groups. Pulmonary sequelae were more serious and recovery was slower in patients with severe/critical disease than in patients with moderate disease. A portion of the fibrosis-like findings were completely absorbed in patients with moderate and severe/critical diseases. KEY POINTS: • Lung sequelae were more serious and recovery was slower in severe/critical COVID-19 patients. • Complete absorption of fibrosis-like findings after a short-term follow-up was observed in at least 17/70 (24.3%) of COVID-19 patients with severe/critical disease and 53/102 (52.0%) of COVID-19 patients with moderate disease. • The most common fibrosis-like findings was a parenchymal band; irregular interface was a nonspecific sign of COVID-19, and the percentage of bronchial dilatation in patients with severe/critical disease remained at a relatively stable medium level (range, 31.6 to 47.8%) at all stages.