ABSTRACT
Global public health will be severely impacted by the successive waves of emerging COVID-19 disease. Since 2019 people get sick and die in our daily lives placing a massive burden on our health system. One of the crucial factors that has led to the virus's fast spread is a protracted clinical testing gap before discovering of a positive or negative result. A detection system based on deep learning was developed by using chest X-ray(CXR) images of Covid19 patient and healthy people. In this regard the Convolution Neural Network along with other DNNs have been proved to produce good results. To improve the COVID-19 detection accuracy, we developed model using the deep learning(CNN) approach where we observed an accuracy of 96%. We validated the accuracy by using same dataset through a pretrained VGG16 model and an LSTM model which produced excellent reliable results. Our aim of this research is to implement a reliable Deep Learning model to detect presence of Covid-19 in case of limited availability of chest-Xray images.
ABSTRACT
Background: Radiomical data are redundant but they might serve as a tool for lung quantitative assessment reflecting disease severity and actual physiological status of COVID-19 patients. Objective: Test the effectiveness of machine learning in eliminating data redundancy of radiomics and reflecting pathophysiologic changes in patients with COVID-19 pneumonia. Methods: We analyzed 605 cases admitted to Al Ain Hospital from 24 February to 1 July, 2020. They met the following inclusion criteria: age $\geq 18$ years;inpatient admission;PCR positive for SARS-CoV-2;lung CT available at PACS. We categorized cases into 4 classes: mild < 5% of pulmonary parenchymal involvement, moderate - 5-24%, severe - 25-49%, and critical $\geq50$ %. We used CT scans to build regression models predicting the oxygenation level, respiratory and cardiovascular functioning. Results: Radiomical findings are a reliable source of information to assess the functional status of patients with COVID-19. Machine learning models can predict the oxygenation level, respiratory and cardiovascular functioning from a set of demographics and radiomics data regardless of the settings of reconstructionkernels. The regression models can be used for scoring lung impairment and comparing disease severity in followup studies. The most accurate prediction we achieved was 6.454 +/- 3.715% of mean absolute error/range for all thefeatures and 7.069 +/- 4.17% for radiomics.Conclusion:The models may contribute to the proper risk evaluation anddisease management especially when the oxygen therapy impacts the actual values of the functional findings. Still,the structural assessment of an acute lung injury reflects the severity of the disease.
ABSTRACT
Background: Radiomical data are redundant but they might serve as a tool for lung quantitative assessment reflecting disease severity and actual physiological status of COVID-19 patients. Objective: Test the effectiveness of machine learning in eliminating data redundancy of radiomics and reflecting pathophysiologic changes in patients with COVID-19 pneumonia. Methods: We analyzed 605 cases admitted to Al Ain Hospital from 24 February to 1 July, 2020. They met the following inclusion criteria: age≥18 years;inpatient admission;PCR positive for SARS-CoV-2;lung CT available at PACS. We categorized cases into 4 classes: mild ≤25% of pulmonary parenchymal involvement, moderate - 25-50%, severe - 50-75%, and critical –over 75%. We used CT scans to build regression models predicting the oxygenation level, respiratory and cardiovascular functioning. Results: Radiomical findings are a reliable source of information to assess the functional status of patients with COVID-19. Machine learning models can predict the oxygenation level, respiratory and cardiovascular functioning from a set of demographics and radiomics data regardless of the settings of reconstruction kernels. The regression models can be used for scoring lung impairment and comparing disease severity in follow up studies. The most accurate prediction we achieved was 6.454±3.715% of mean absolute error/range for all the features and 7.069±4.17% for radiomics. Conclusion: The models may contribute to the proper risk evaluation and disease management especially when the oxygen therapy impacts the actual values of the functional findings. Still, the structural assessment of an acute lung injury reflects the severity of the disease. Author
ABSTRACT
Background: Chest radiographs are frequently used to evaluate pediatric patients with COVID-19 infection during the current pandemic. Despite the minimal radiation dose associated with chest radiography, children are far more sensitive to ionizing radiation's carcinogenic effects than adults. This study aimed to examine whether serum biochemical markers could be potentially used as a surrogate for imaging findings to reduce radiation exposure. Methods: The retrospective posthoc analysis of 187 pediatric patients who underwent initial chest radiographs and serum biochemical parameters on the first day of emergency department admission. The cohort was separated into two groups according to whether or not the initial chest radiograph revealed evidence of pneumonia. Spearman's rank correlation was used to connect serum biochemical markers with observations on chest radiographs. The Student's t-test was employed for normally distributed data, and for non-normally distributed data, the Mann–Whitney U test was used. A simple binary logistic regression was used to determine the importance of LDH in predicting chest radiographs. The discriminating ability of LDH in predicting chest radiographs was determined using receiver operating characteristics (ROC) analysis. The cut-off value was determined using Youden's test. Interobserver agreement was quantified using the Cohen k coefficient. Results: 187 chest radiographs from 187 individual pediatric patients (95 boys and 92 girls;mean age ± SD, 10.1 ± 6.0 years;range, nine months–18 years) were evaluated. The first group has 103 patients who did not have pneumonia on chest radiographs, while the second group contains 84 patients who had evidence of pneumonia on chest radiographs. GGO, GGO with consolidation, consolidation, and peri-bronchial thickening were deemed radiographic evidence of pneumonia in group 2 patients. Individuals in group 2 with radiological indications of pneumonia had significantly higher LDH levels (p = 0.001) than patients in group 1. The Spearman's rank correlation coefficient between LDH and chest radiography score is 0.425, showing a significant link. With a p-value of < 0.001, the simple binary logistic regression analysis result validated the relevance of LDH in predicting chest radiography. An abnormal chest radiograph was related to LDH > 200.50 U/L (AUC = 0.75), according to the ROC method. Interobserver agreement between the two reviewers was almost perfect for chest radiography results in both groups (k = 0.96, p = 0.001). Conclusion: This study results show that, compared to other biochemical indicators, LDH has an 80.6% sensitivity and a 62% specificity for predicting abnormal chest radiographs in a pediatric patient with confirmed COVID-19 infection. It also emphasizes that biochemical measures, rather than chest radiological imaging, can detect the pathogenic response to COVID-19 infection in the chest earlier. As a result, we hypothesized LDH levels might be potentially used instead of chest radiography in children with COVID-19, reducing radiation exposure.
ABSTRACT
Background: Coronavirus disease 2019 (COVID-19) caused by SARS-CoV-2 was first reported in Wuhan in 2019 and reached pandemic proportions. SARS-CoV-2-related respiratory failure and acute kidney injury (AKI) are major complications of infection. Kidney Injury Molecule-1 (KIM-1) is a scavenger receptor expressed by kidney epithelial cells and was previously reported to be a receptor for Hepatitis virus A. We hypothesized that KIM-1 is a receptor for SARS-CoV-2 and may play an important role in COVID-19 lung and kidney injury. Methods: Human lung and kidney autopsy samples were immunostained and analyzed. Liposomal nanoparticles displaying the SARS-CoV-2 spike protein on their surface (virosomes) were generated. Virosome uptake by A549 lung epithelial cells, mouse primary lung epithelial cells and human kidney tubuloids (3D structures of kidney epithelial cells) was evaluated in the presences or absence of anti-KIM-1 antibody or TW-37, a small molecule inhibitor of KIM-1-mediated endocytosis that we discovered. Protein-protein interaction characteristics between purified SARS-CoV-2 spike protein and purified KIM-1 were determined using microscale thermophoresis. HEK293 cells expressing human KIM-1 but not angiotensin-converting enzyme 2 (ACE2) were infected with live SARS-CoV-2 or pseudovirions expressing the SARS-CoV-2 spike protein. Results: KIM-1 was expressed in lung and kidney epithelial cells in COVID-19 patient autopsy samples. Human and mouse lung and kidney epithelial cells expressed KIM-1 and endocytosed spike-virosomes. Both anti-KIM-1 antibodies and TW-37 inhibited uptake. Enhanced KIM-1 expression by human kidney tubuloids increased virosome uptake. KIM-1 positive cells expressed less ACE2. Using microscale thermophoresis, the EC50 for interaction between KIM-1 and SARS-CoV-2 spike protein and the receptor binding domain were 56.2±28.8 nM and 9.95±3.10 nM, respectively. KIM-1-expressing HEK293 cells without ACE2 expression had increased susceptibility to infection by live SARS-CoV-2 and pseudovirions expressing spike when compared with control cells. Conclusions: KIM-1 is a receptor for SARS-CoV-2 in the lung and kidney and thus, KIM-1 inhibitors such as TW-37 can be potential therapeutics and/or prophylactic agents for COVID-19.
ABSTRACT
Background: Despite the dominance of Covid-19 in the current situation, MERS-CoV is found infrequently in the Middle East. When coupled with the chest radiographic score, serum biochemical parameters may be utilized to assess serum biochemical changes in individuals with different degrees of MERS-CoV infection and to predict death. The purpose of this study was to examine the association between increased LDH levels and severe MERS-CoV outcomes utilizing ventilation days and an elevated chest radiographic score. Results: Fifty-seven patients were included in the retrospective cohort. The mean age was 44.9 ± 13.5 years, while the range was between 12 and 73 years. With an average age of 53.3 ± 16.5 years, 18 of 57 (31.6%) patients were classified as deceased. The deceased group showed a substantially greater amount of LDH than the recovery group (280.18 ± 150.79 vs. 1241.72 ± 1327.77, p = 0.007). A cut-off value of > 512 LDH was established with a C-statistic of 0.96 (95% CI 0.92–1.00) and was 94% sensitive and 93% specific for mortality. Multivariate cox regression analysis revealed that loge (LDH) (adjusted HR: 9.91, 95% CI: 2.44–40.3, p = 0.001) and chest radiographic score (adjusted HR: 1.24, 95% CI: 1.05–1.47, p = 0.01) were risk factors for mortality, whereas ventilation days were a protective factor (adjusted HR: 0.84, 95% CI: 0.76–0.93, p = 0.001). Conclusion: According to our results, blood LDH levels of > 512 had a 94% sensitivity and 93% specificity for predicting in-hospital mortality in patients infected with MERS-CoV. The chest radiographic score of 11.34 ± 5.4 was the risk factor for the mortality (adjusted Hazard ratio HR: 1.24, 95% CI: 1.05–1.47, p = 0.01). Thus, threshold may aid in the identification of individuals with MERS-CoV infection who die in hospital.