Automatic detection of COVID-19 and community-acquired pneumonia on CT images with artificial intelligence (preprint)

The outbreak of COVID-19 plunged the world into unprecedented difficulties, bringing life to a terrible standstill around the world and claiming thousands of lives. It seriously endangers human life and health. Thus, a high-efficiency COVID-19 and community-acquired pneumonia (CAP) detection method based on computed tomography (CT) images were developed to combat the pandemic. Sixty-one confirmed COVID-19 cases were consecutively enrolled together with 27 confirmed community-acquired pneumonia patients and 87 normal control patients with no finding of COVID-19 and CAP from Ruian People’s Hospital, from January 2020 to March 2020. The final cohort of 175 patients (1611 CT slices) was divided into training and test sets in an 8:2 and equal ratio. Three senior radiologists assessed clinical Characteristics of COVID-19 patients and abnormalities in chest CT images independently. The proposed model, EfficientNet-B1 was applied for the classification task, which consists of 16 mobile inverted bottleneck convolutions (MBConv), 2 convolutional (Conv) layers, 1 global average-pooling layer, and 1 fully connected layer, obtained a test accuracy of 94.24% and 93.64% for binary and three-class classification, respectively. It also achieved a sensitivity of 92.35%, a specificity of 95.93%, a positive predictive value of 95.26%, negative predictive value of 93.39% with an area under the receiver operating characteristic curve of 0.9414 for COVID-19 and CAP classification. Fever (58 [95.1%]) with a mean axillary temperature of 37.3 ± 0.8 °C, cough (52 [85.2%]), and sputum production (35 [57.4%]) are the most common symptoms of COVID-19 patients at the onset of illness. More importantly, more than 50% of COVID-19 images show ground-glass opacities. The proposed model achieves state-of-the-art performance and can be applied as a supplementary tool for radiologists, which is beneficial for pandemic control. 

Clinical Characteristics and Outcomes in Patients with COVID-19 Received Tocilizumab in Wuhan, China: A Multicentre Retrospective Cohort Study (preprint)

Background: Coronavirus Disease-2019 (COVID-19) has caused considerable morbidity and mortality. Hence, there is an urgency to find effective treatment. Tocilizumab, an inhibitor of IL-6, has been widely proposed as a treatment of severely ill patients without robust evidence supporting its use. Methods: In this multicentre, retrospective, cohort study, we included 5,235 adult patients who were admitted to 3 hospitals in Wuhan, China with confirmed COVID-19 from January 20 to March 18, 2020 . 65 patients in tocilizumab group and 130 patients in non-tocilizumab group were propensity score matched at a ratio of 2:1 based on age, sex, and comorbidities. Detailed demographic data, comorbidities, radiological and laboratory parameters, complications and treatments were compared between tocilizumab group and non-tocilizumab group. Furthermore, univariable and multivariable Logistic and Cox regression models were used to explore the risk of complications and in-hospital death associated with tocilizumab. Findings: During the follow-up, patients in non-tocilizumab group were more likely to develop into death (42 [32·31%] vs 14 [21·54%]). After adjusting for confounding, the detected risk for in-hospital death was lower in the tocilizumab group versus the non-tocilizumab group (HR=0·47; 95% CI=0·25-0·90; p=0·023). In the multivariable logistic regression model, use of tocilizumab was associated with a lower risk of ARDS (OR=0 · 23; 95% CI=0·11-0·45; p<0·0001). Before treatment the patients had heightened inflammation and more dysregulated immune cells, which might aggravate disease progression. However, abnormally elevated IL-6, CRP, fibrinogen and APTT decreased in COVID-19 patients after treatment. And the counts of lymphocytes and immune cells subset in peripheral blood, which decreased in patients, returned to normal after treatment. No obvious complications were observed. Interpretation: Tocilizumab may be of value in improving outcomes in severe patients of COVID-19, which provided a novel strategy for COVID-19-induced cytokine release syndrome (CRS). Our preliminary data could inform bedside decisions until more data from randomized, controlled clinical trials becomes available.Funding Statement: SARS-CoV-2 Pneumonia Emergency Technology Public Relations Project of Tongji Medical College, Huazhong University of Science and Technology (No. 2020kfyXGYJ043) and National Key Research and Development Plan for the Emergency Management of Novel Coronavi rus Pneumonia, China (No. 2020YFC0845100).Declaration of Interests: The authors report no conflicts of interest.Ethics Approval Statement: This study was approved by the Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology (TJ-C20200108) and granted a waiver of informed consent from study participants.

Differentiating novel coronavirus pneumonia from general pneumonia based on machine learning (preprint)

Background: Chest CT screening as supplementary means is crucial in diagnosing novel coronavirus pneumonia (COVID-19) with high sensitivity and popularity. Machine learning was adept in discovering intricate structures from CT images and achieved expert-level performance in medical image analysis. Methods: : An integrated machine learning framework on chest CT images for differentiating COVID-19 from general pneumonia (GP) was developed and validated. Seventy-three confirmed COVID-19 cases were consecutively enrolled together with twenty-seven confirmed general pneumonia patients from Ruian People’s Hospital, from January 2020 to March 2020. To accurately classify COVID-19, region of interest (ROI) delineation was implemented based on ground glass opacities (GGOs) before feature extraction. Then, 34 statistical texture features of COVID-19 and GP ROI images were extracted, including 13 gray level co-occurrence matrix (GLCM) features, 15 gray level-gradient co-occurrence matrix (GLGCM) features and 6 histogram features. High dimensional features impact the classification performance. Thus, ReliefF algorithm was leveraged to select features. The relevance of each features was the average weights calculated by ReliefF in n times. Features with relevance lager than the empirically set threshold T were selected. After feature selection, the optimal feature set along with 4 other selected feature combinations for comparison were applied to the ensemble of bagged tree (EBT) and four other machine learning classifiers including support vector machine (SVM), logistic regression (LR), decision tree (DT), and K-nearest neighbor with Minkowski distance equal weight (KNN) using 10-fold cross-validation. Results: and Conclusions: The classification accuracy (ACC), sensitivity (SEN), specificity (SPE) of our proposed method yield 94.16%, 88.62% and 100.00%, respectively. The area under the receiver operating characteristic curve (AUC) was 0.99. The experimental results indicate that the EBT algorithm with statistical textural features based on GGOs for differentiating COVID-19 from general pneumonia achieved high transferability, efficiency, specificity, sensitivity, and impressive accuracy, which is beneficial for inexperienced doctors to more accurately diagnose COVID-19 and essential for controlling the spread of the disease.

Membrane Nanoparticles Derived from ACE2-rich Cells Block SARS-CoV-2 Infection (preprint)

The ongoing COVID-19 epidemic worldwide necessitates the development of novel effective agents against SARS-CoV-2. ACE2 is the main receptor of SARS-CoV-2 S1 protein and mediates viral entry into host cells. Herein, the membrane nanoparticles prepared from ACE2-rich cells are discovered with potent capacity to block SARS-CoV-2 infection. The membrane of human embryonic kidney-239T cell highly expressing ACE2 is screened to prepare nanoparticles. The nanomaterial termed HEK-293T-hACE2 NPs contains 265.1 ng mg-1 of ACE2 on the surface and acts as a bait to trap SARS-CoV-2 S1 in a dose-dependent manner, resulting in reduced recruitment of the viral ligand to host cells. Interestingly, SARS-CoV-2 S1 can translocate to the cytoplasm and affect the cell metabolism, which is also inhibited by HEK-293T-hACE2 NPs. Further studies reveal that HEK-293T-hACE2 NPs can efficiently suppress SARS-CoV-2 S pseudovirions entry into human proximal tubular cells and block viral infection with a low half maximal inhibitory concentration. Additionally, this biocompatible membrane nanomaterial is sufficient to block the adherence of SARS-CoV-2 D614G-S1 mutant to sensitive cells. Our study demonstrates a easy-to-achieve membrane nano-antagonist for curbing SARS-CoV-2, which enriches the existing antiviral arsenal and provides new possibilities to treat COVID-19.

CT findings of patients infected with SARS-CoV-2 (preprint)

Objective: We aimed to describe the chest CT findings in sixty-seven patients infected by Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Method and material: We retrospectively reviewed 67 patients hospitalized in Ruian People's Hospital. All the patients received the positive diagnosis of SARS-CoV-2 infection. The CT and clinical data were collected between January 23 rd , 2020 and February 10 th , 2020. The CT images were analyzed by the senior radiologists. Conclusion: There are 54 patients with positive CT findings and 13 patients with negative CT findings. The typical CT findings in hospitalized patients with SARS-CoV-2 infection were ground glass opacities (42/54), lesions located in the peripheral area (50/54), multiple lesions (46/54), and lesions located in the lower lobes (42/54). There were less typical CT findings, including air bronchogram (18/54), pleural thickening or pleural effusion (14/54), consolidation (12/54), lesions in the upper lobes (12/54), interlobular septal thickening (11/54), reversed halo sign (9/54), single lesion (8/54), air cavities (4/54), bronchial wall thickening (3/54), and intrathoracic lymph node enlargement (2/54).

Clinical and Radiographic Characteristics, Management and Short-term Outcomes of Patients with COVID-19 in Wenzhou, China (preprint)

Background: Coronavirus disease 2019 (COVID-19) is an emerging viral disease. Here, we reported the clinical features, management, and short-term outcomes of COVID-19 patients in Wenzhou, an area outside Wuhan.Methods: Patients admitted to the Infectious Diseases Department of Ruian People's Hospital in Wenzhou, from January 21 to February 7, 2020, were recruited. Medical data on epidemiological history, demographics, clinical characteristics, laboratory tests, computerized tomography (CT) examination, treatment, and short-term outcomes were retrospectively reviewed. Blood biochemistry and routine tests were examined using standard methods and automatic machines. CT examination was performed again for several times during the hospitalization as necessary.Results: A total of 67 confirmed COVID-19 cases were diagnosed; 64 (95.4%) were common cases and three (4.5%) severe cases. The most common symptoms at admission were fever (86.6%), cough (77.6%), productive cough (52.2%), chest distress (17.9%), and sore throat (11.9%), followed by diarrhea (7.4%), headache (7.4%), shortness of breath (6.0%), dizziness (4.5%), muscular soreness (4.5%), and running nose (4.5%). Thirty patients (47.8%) had increased C-reactive protein levels. The CT radiographs at admission showed abnormal findings in 54 (80.6%) patients. The patients were treated mainly by oxygen therapy and antiviral drugs. By February 17, 2020, none of the 67 patients died and no infection occurred among medical staff in the department. Fifty-four (80.6%) patients were completely recovered and all others were improving.Conclusion: Cases in Wenzhou are mild, with good prognosis. Timely and appropriate screening, diagnosis, and treatment are the key to achieve the good outcomes.