Your browser doesn't support javascript.
Show: 20 | 50 | 100
Results 1 - 20 de 185
Filter
1.
Comput Intell Neurosci ; 2022: 6185013, 2022.
Article in English | MEDLINE | ID: covidwho-1861702

ABSTRACT

It is critical to establish a reliable method for detecting people infected with COVID-19 since the pandemic has numerous harmful consequences worldwide. If the patient is infected with COVID-19, a chest X-ray can be used to determine this. In this work, an X-ray showing a COVID-19 infection is classified by the capsule neural network model we trained to recognise. 6310 chest X-ray pictures were used to train the models, separated into three categories: normal, pneumonia, and COVID-19. This work is considered an improved deep learning model for the classification of COVID-19 disease through X-ray images. Viewpoint invariance, fewer parameters, and better generalisation are some of the advantages of CapsNet compared with the classic convolutional neural network (CNN) models. The proposed model has achieved an accuracy greater than 95% during the model's training, which is better than the other state-of-the-art algorithms. Furthermore, to aid in detecting COVID-19 in a chest X-ray, the model could provide extra information.


Subject(s)
COVID-19 , Deep Learning , COVID-19/diagnostic imaging , Humans , Neural Networks, Computer , Thorax/diagnostic imaging , X-Rays
2.
Respir Investig ; 60(3): 364-368, 2022 May.
Article in English | MEDLINE | ID: covidwho-1805070

ABSTRACT

BACKGROUND: Because of genetic mutations occurring during viral replication, new SARS-CoV-2 variants will continue to emerge. Throughout the COVID-19 pandemic, thorax computed tomographic (CT) findings have played a crucial role in the diagnosis and follow-up of patients with COVID-19. In this study, we compared the thorax CT findings of patients infected with SARS-CoV-2 variants (variant group) with those of patients infected with the non-variant strain (non-variant group) to assess if thorax CT findings may be utilized to discriminate between the groups. Furthermore, we compared demographic and laboratory data between the groups. METHODS: The study comprised a total of 77 patients who presented to our hospital with a preliminary diagnosis of COVID-19 based on clinical symptoms, a positive oropharyngeal/nasopharyngeal swab RT-PCR testing, and thorax CT examinations. Patients' laboratory and demographic features as well as thorax CT findings were retrospectively evaluated, and the results were grouped according to RT-PCR results. RESULTS: There were 42 patients in the non-variant group and 35 patients in the variant group. The average age of patients infected with the non-variant strain, alpha variant, and gamma variant was 63.52 ± 14.87 years, 54.86 ± 14.31 years, and 59.4 ± 17.79 years, respectively. The average age of the variant group was significantly lower than that of the non-variant group. There was no significant difference in thorax CT findings between the groups, and consolidation, ground glass densities, and cobblestone pattern in the bilateral lower lobes and peripheral areas were the most common thorax CT findings in both the groups. CONCLUSION: There is no significant difference in thorax CT findings between the variant and non-variant groups. Therefore, clinical and laboratory characteristics should take precedence over thorax CT findings for distinguishing between patients infected with SARS-CoV-2 variants and the non-variant strain.


Subject(s)
COVID-19 , SARS-CoV-2 , Aged , COVID-19/diagnostic imaging , Humans , Lung , Middle Aged , Pandemics , Retrospective Studies , SARS-CoV-2/genetics , Thorax/diagnostic imaging , Tomography, X-Ray Computed/methods
3.
PLoS One ; 17(3): e0264711, 2022.
Article in English | MEDLINE | ID: covidwho-1793510

ABSTRACT

Reports detailing the clinical characteristics, viral load, and outcomes of patients with normal initial chest CT findings are lacking. We sought to compare the differences in clinical findings, viral loads, and outcomes between patients with confirmed COVID-19 who initially tested negative on chest CT (CT negative) with patients who tested initially positive on chest CT (CT positive). The clinical data, viral loads, and outcomes of initial CT-positive and CT-negative patients examined between January 2020 and April 2020 were retrospectively compared. The efficacy of viral load (cyclic threshold value [Ct value]) in predicting pneumonia was evaluated using receiver operating characteristic (ROC) curve and area under the curve (AUC). In total, 128 patients underwent initial chest CT (mean age, 54.3 ± 19.0 years, 50% male). Of those, 36 were initially CT negative, and 92 were CT positive. The CT-positive patients were significantly older (P < .001) than the CT-negative patients. Only age was significantly associated with the initial presence of pneumonia (odds ratio, 1.060; confidence interval (CI), 1.020-1-102; P = .003). In addition, age (OR, 1.062; CI, 1.014-1.112; P = .011), fever at diagnosis (OR, 6.689; CI, 1.715-26.096; P = .006), and CRP level (OR, 1.393; CI, 1.150-1.687; P = .001) were significantly associated with the need for O2 therapy. Viral load was significantly higher in the CT-positive group than in the CT-negative group (P = .017). The cutoff Ct value for predicting the presence of pneumonia was 27.71. Outcomes including the mean hospital stay, intensive care unit admission, and O2 therapy were significantly worse in the CT-positive group than in the CT-negative group (all P < .05). In conclusion, initially CT-negative patients showed better outcomes than initially CT-positive patients. Age was significantly associated with the initial presence of pneumonia, and viral load may help in predicting the initial presence of pneumonia.


Subject(s)
COVID-19/diagnosis , Thorax/diagnostic imaging , Viral Load , Adult , Aged , COVID-19/epidemiology , COVID-19/virology , Female , Humans , Male , Middle Aged , Predictive Value of Tests , Prognosis , Republic of Korea/epidemiology , Retrospective Studies , SARS-CoV-2 , Sputum/virology , Tomography, X-Ray Computed , Viral Load/physiology , Young Adult
4.
Phys Med Biol ; 67(7)2022 03 29.
Article in English | MEDLINE | ID: covidwho-1774310

ABSTRACT

Chest x-ray (CXR) is one of the most commonly used imaging techniques for the detection and diagnosis of pulmonary diseases. One critical component in many computer-aided systems, for either detection or diagnosis in digital CXR, is the accurate segmentation of the lung. Due to low-intensity contrast around lung boundary and large inter-subject variance, it has been challenging to segment lung from structural CXR images accurately. In this work, we propose an automatic Hybrid Segmentation Network (H-SegNet) for lung segmentation on CXR. The proposed H-SegNet consists of two key steps: (1) an image preprocessing step based on a deep learning model to automatically extract coarse lung contours; (2) a refinement step to fine-tune the coarse segmentation results based on an improved principal curve-based method coupled with an improved machine learning method. Experimental results on several public datasets show that the proposed method achieves superior segmentation results in lung CXRs, compared with several state-of-the-art methods.


Subject(s)
Lung Diseases , Neural Networks, Computer , Humans , Image Processing, Computer-Assisted/methods , Lung/diagnostic imaging , Lung Diseases/diagnosis , Radiography , Thorax/diagnostic imaging
5.
PLoS One ; 17(2): e0264172, 2022.
Article in English | MEDLINE | ID: covidwho-1708123

ABSTRACT

During the SARS-CoV-2 pandemic, chest X-Ray (CXR) scores are essential to rapidly assess patients' prognoses. This study evaluates a published CXR score in a different national healthcare system. In our study, this CXR score maintains a prognostic role in predicting length of hospital stay, but not disease severity. However, our results show that the predictive role of CXR score could be influenced by socioeconomic status and healthcare system.


Subject(s)
COVID-19/pathology , Thorax/diagnostic imaging , Adult , Body Mass Index , COVID-19/virology , Comorbidity , Female , Humans , Length of Stay , Male , Middle Aged , Prognosis , Radiography, Thoracic , Retrospective Studies , SARS-CoV-2/isolation & purification , Severity of Illness Index , Smoking
6.
Biomark Med ; 16(4): 291-301, 2022 03.
Article in English | MEDLINE | ID: covidwho-1706742

ABSTRACT

Aim: Pulmonary disease burden and biomarkers are possible predictors of outcomes in patients with COVID-19 and provide complementary information. In this study, the prognostic value of adding quantitative chest computed tomography (CT) to a multiple biomarker approach was evaluated among 148 hospitalized patients with confirmed COVID-19. Materials & methods: Patients admitted between March and July 2020 who were submitted to chest CT and biomarker measurement (troponin I, D-dimer and C-reactive protein) were retrospectively analyzed. Biomarker and tomographic data were compared and associated with death and intensive care unit admission. Results: The number of elevated biomarkers was significantly associated with greater opacification percentages, lower lung volumes and higher death and intensive care unit admission rates. Total lung volume <3.0 l provided further stratification for mortality when combined with biomarker evaluation. Conclusion: Adding automated CT data to a multiple biomarker approach may provide a simple strategy for enhancing risk stratification of patients with COVID-19.


Subject(s)
Biomarkers/analysis , COVID-19/diagnosis , Thorax/diagnostic imaging , Aged , Aged, 80 and over , Biomarkers/blood , C-Reactive Protein/analysis , COVID-19/mortality , COVID-19/virology , Female , Fibrin Fibrinogen Degradation Products/analysis , Hospital Mortality , Humans , Intensive Care Units , Length of Stay , Male , Middle Aged , Prognosis , Retrospective Studies , Risk Factors , SARS-CoV-2/genetics , SARS-CoV-2/isolation & purification , Tomography, X-Ray Computed , Troponin I/blood
7.
PLoS One ; 17(2): e0263922, 2022.
Article in English | MEDLINE | ID: covidwho-1686110

ABSTRACT

IMPORTANCE: When hospitals are at capacity, accurate deterioration indices could help identify low-risk patients as potential candidates for home care programs and alleviate hospital strain. To date, many existing deterioration indices are based entirely on structured data from the electronic health record (EHR) and ignore potentially useful information from other sources. OBJECTIVE: To improve the accuracy of existing deterioration indices by incorporating unstructured imaging data from chest radiographs. DESIGN, SETTING, AND PARTICIPANTS: Machine learning models were trained to predict deterioration of patients hospitalized with acute dyspnea using existing deterioration index scores and chest radiographs. Models were trained on hospitalized patients without coronavirus disease 2019 (COVID-19) and then subsequently tested on patients with COVID-19 between January 2020 and December 2020 at a single tertiary care center who had at least one radiograph taken within 48 hours of hospital admission. MAIN OUTCOMES AND MEASURES: Patient deterioration was defined as the need for invasive or non-invasive mechanical ventilation, heated high flow nasal cannula, IV vasopressor administration or in-hospital mortality at any time following admission. The EPIC deterioration index was augmented with unstructured data from chest radiographs to predict risk of deterioration. We compared discriminative performance of the models with and without incorporating chest radiographs using area under the receiver operating curve (AUROC), focusing on comparing the fraction and total patients identified as low risk at different negative predictive values (NPV). RESULTS: Data from 6278 hospitalizations were analyzed, including 5562 hospitalizations without COVID-19 (training cohort) and 716 with COVID-19 (216 in validation, 500 in held-out test cohort). At a NPV of 0.95, the best-performing image-augmented deterioration index identified 49 more (9.8%) individuals as low-risk compared to the deterioration index based on clinical data alone in the first 48 hours of admission. At a NPV of 0.9, the EPIC image-augmented deterioration index identified 26 more individuals (5.2%) as low-risk compared to the deterioration index based on clinical data alone in the first 48 hours of admission. CONCLUSION AND RELEVANCE: Augmenting existing deterioration indices with chest radiographs results in better identification of low-risk patients. The model augmentation strategy could be used in the future to incorporate other forms of unstructured data into existing disease models.


Subject(s)
Clinical Deterioration , Thorax/diagnostic imaging , Adolescent , Adult , Aged , Aged, 80 and over , COVID-19/pathology , COVID-19/virology , Dyspnea/pathology , Female , Hospitalization , Humans , Machine Learning , Male , Middle Aged , ROC Curve , Respiration, Artificial , Retrospective Studies , Risk Factors , SARS-CoV-2/isolation & purification , Young Adult
8.
Sci Rep ; 12(1): 1716, 2022 02 02.
Article in English | MEDLINE | ID: covidwho-1665719

ABSTRACT

The rapid evolution of the novel coronavirus disease (COVID-19) pandemic has resulted in an urgent need for effective clinical tools to reduce transmission and manage severe illness. Numerous teams are quickly developing artificial intelligence approaches to these problems, including using deep learning to predict COVID-19 diagnosis and prognosis from chest computed tomography (CT) imaging data. In this work, we assess the value of aggregated chest CT data for COVID-19 prognosis compared to clinical metadata alone. We develop a novel patient-level algorithm to aggregate the chest CT volume into a 2D representation that can be easily integrated with clinical metadata to distinguish COVID-19 pneumonia from chest CT volumes from healthy participants and participants with other viral pneumonia. Furthermore, we present a multitask model for joint segmentation of different classes of pulmonary lesions present in COVID-19 infected lungs that can outperform individual segmentation models for each task. We directly compare this multitask segmentation approach to combining feature-agnostic volumetric CT classification feature maps with clinical metadata for predicting mortality. We show that the combination of features derived from the chest CT volumes improve the AUC performance to 0.80 from the 0.52 obtained by using patients' clinical data alone. These approaches enable the automated extraction of clinically relevant features from chest CT volumes for risk stratification of COVID-19 patients.


Subject(s)
COVID-19/diagnosis , COVID-19/virology , Deep Learning , SARS-CoV-2 , Thorax/diagnostic imaging , Thorax/pathology , Tomography, X-Ray Computed , Algorithms , COVID-19/mortality , Databases, Genetic , Humans , Image Interpretation, Computer-Assisted/methods , Image Processing, Computer-Assisted/methods , Prognosis , Tomography, X-Ray Computed/methods , Tomography, X-Ray Computed/standards
9.
Eur Rev Med Pharmacol Sci ; 26(1): 298-304, 2022 01.
Article in English | MEDLINE | ID: covidwho-1633445

ABSTRACT

OBJECTIVE: The novel coronavirus disease 2019 (COVID-19) may affect the adrenal glands. Therefore, it is important to evaluate the morphologic appearance of the adrenal glands by thorax computed tomography (CT). On CT scans, stranding in peripheral fatty tissue with enlarged adrenal glands may indicate signs of adrenal infarction (SAI). The present study aimed to evaluate the incidence of SAI and determine whether this finding may contribute to predictions of the prognosis of COVID-19. PATIENTS AND METHODS: A total of 343 patients who had been hospitalized at Malatya Training and Research Hospital between September 1 and 30, 2020, with a diagnosis of COVID-19 were enrolled in this study. All patients underwent thorax CT scans that included their adrenal glands. RESULTS: Of the enrolled patients, 16.0% had SAI. Moreover, 41.8% of patients with SAI and 15.3% of patients without SAI were treated in the Intensive Care Unit (ICU). Patients with SAI had a significantly higher rate of ICU admission (p < 0.001). Mortality rates were also significantly higher among patients with SAI than those without p < 0.001). CONCLUSIONS: In this study, it was found that COVID-19 patients with SAI may have a poorer prognosis. More comprehensive studies are needed on this subject, but the present study may provide helpful preliminary information in terms of prognosis.


Subject(s)
Adrenal Gland Diseases/diagnostic imaging , Adrenal Glands/diagnostic imaging , COVID-19/diagnosis , Adrenal Gland Diseases/etiology , Aged , Aged, 80 and over , COVID-19/complications , COVID-19/mortality , Female , Hospitalization , Humans , Intensive Care Units , Logistic Models , Male , Middle Aged , Prognosis , Retrospective Studies , Thorax/diagnostic imaging , Tomography, X-Ray Computed
10.
Sci Rep ; 12(1): 815, 2022 01 17.
Article in English | MEDLINE | ID: covidwho-1629950

ABSTRACT

Deep neural networks (DNNs) have been widely applied for detecting COVID-19 in medical images. Existing studies mainly apply transfer learning and other data representation strategies to generate accurate point estimates. The generalization power of these networks is always questionable due to being developed using small datasets and failing to report their predictive confidence. Quantifying uncertainties associated with DNN predictions is a prerequisite for their trusted deployment in medical settings. Here we apply and evaluate three uncertainty quantification techniques for COVID-19 detection using chest X-Ray (CXR) images. The novel concept of uncertainty confusion matrix is proposed and new performance metrics for the objective evaluation of uncertainty estimates are introduced. Through comprehensive experiments, it is shown that networks pertained on CXR images outperform networks pretrained on natural image datasets such as ImageNet. Qualitatively and quantitatively evaluations also reveal that the predictive uncertainty estimates are statistically higher for erroneous predictions than correct predictions. Accordingly, uncertainty quantification methods are capable of flagging risky predictions with high uncertainty estimates. We also observe that ensemble methods more reliably capture uncertainties during the inference. DNN-based solutions for COVID-19 detection have been mainly proposed without any principled mechanism for risk mitigation. Previous studies have mainly focused on on generating single-valued predictions using pretrained DNNs. In this paper, we comprehensively apply and comparatively evaluate three uncertainty quantification techniques for COVID-19 detection using chest X-Ray images. The novel concept of uncertainty confusion matrix is proposed and new performance metrics for the objective evaluation of uncertainty estimates are introduced for the first time. Using these new uncertainty performance metrics, we quantitatively demonstrate when we could trust DNN predictions for COVID-19 detection from chest X-rays. It is important to note the proposed novel uncertainty evaluation metrics are generic and could be applied for evaluation of probabilistic forecasts in all classification problems.


Subject(s)
COVID-19/diagnostic imaging , Deep Learning , Thorax/diagnostic imaging , Humans
11.
PLoS One ; 16(12): e0261307, 2021.
Article in English | MEDLINE | ID: covidwho-1598199

ABSTRACT

Medical images commonly exhibit multiple abnormalities. Predicting them requires multi-class classifiers whose training and desired reliable performance can be affected by a combination of factors, such as, dataset size, data source, distribution, and the loss function used to train deep neural networks. Currently, the cross-entropy loss remains the de-facto loss function for training deep learning classifiers. This loss function, however, asserts equal learning from all classes, leading to a bias toward the majority class. Although the choice of the loss function impacts model performance, to the best of our knowledge, we observed that no literature exists that performs a comprehensive analysis and selection of an appropriate loss function toward the classification task under study. In this work, we benchmark various state-of-the-art loss functions, critically analyze model performance, and propose improved loss functions for a multi-class classification task. We select a pediatric chest X-ray (CXR) dataset that includes images with no abnormality (normal), and those exhibiting manifestations consistent with bacterial and viral pneumonia. We construct prediction-level and model-level ensembles to improve classification performance. Our results show that compared to the individual models and the state-of-the-art literature, the weighted averaging of the predictions for top-3 and top-5 model-level ensembles delivered significantly superior classification performance (p < 0.05) in terms of MCC (0.9068, 95% confidence interval (0.8839, 0.9297)) metric. Finally, we performed localization studies to interpret model behavior and confirm that the individual models and ensembles learned task-specific features and highlighted disease-specific regions of interest. The code is available at https://github.com/sivaramakrishnan-rajaraman/multiloss_ensemble_models.


Subject(s)
Algorithms , Diagnostic Imaging , Image Processing, Computer-Assisted/classification , Area Under Curve , Entropy , Humans , Lung/diagnostic imaging , ROC Curve , Thorax/diagnostic imaging , X-Rays
12.
PLoS One ; 16(3): e0247686, 2021.
Article in English | MEDLINE | ID: covidwho-1574773

ABSTRACT

OBJECTIVES: The aim of this study was to investigate possible patterns of demand for chest imaging during the first wave of the SARS-CoV-2 pandemic and derive a decision aid for the allocation of resources in future pandemic challenges. MATERIALS AND METHODS: Time data of requests for patients with suspected or confirmed coronavirus disease 2019 (COVID-19) lung disease were analyzed between February 27th and May 27th 2020. A multinomial logistic regression model was used to evaluate differences in the number of requests between 3 time intervals (I1: 6am - 2pm, I2: 2pm - 10pm, I3: 10pm - 6am). A cosinor model was applied to investigate the demand per hour. Requests per day were compared to the number of regional COVID-19 cases. RESULTS: 551 COVID-19 related chest imagings (32.8% outpatients, 67.2% in-patients) of 243 patients were conducted (33.3% female, 66.7% male, mean age 60 ± 17 years). Most exams for outpatients were required during I2 (I1 vs. I2: odds ratio (OR) = 0.73, 95% confidence interval (CI) 0.62-0.86, p = 0.01; I2 vs. I3: OR = 1.24, 95% CI 1.04-1.48, p = 0.03) with an acrophase at 7:29 pm. Requests for in-patients decreased from I1 to I3 (I1 vs. I2: OR = 1.24, 95% CI 1.09-1.41, p = 0.01; I2 vs. I3: OR = 1.16, 95% CI 1.05-1.28, p = 0.01) with an acrophase at 12:51 pm. The number of requests per day for outpatients developed similarly to regional cases while demand for in-patients increased later and persisted longer. CONCLUSIONS: The demand for COVID-19 related chest imaging displayed distinct distribution patterns depending on the sector of patient care and point of time during the SARS-CoV-2 pandemic. These patterns should be considered in the allocation of resources in future pandemic challenges with similar disease characteristics.


Subject(s)
COVID-19/diagnostic imaging , Diagnostic Imaging/trends , Thorax/diagnostic imaging , Adult , Aged , COVID-19/epidemiology , Diagnostic Tests, Routine/trends , Female , Humans , Male , Middle Aged , Models, Theoretical , Pandemics , Pilot Projects , SARS-CoV-2/pathogenicity , Thorax/virology
13.
Viruses ; 13(12)2021 11 26.
Article in English | MEDLINE | ID: covidwho-1542798

ABSTRACT

Overactivation of the complement system has been characterized in severe COVID-19 cases. Complement components are known to trigger NETosis via the coagulation cascade and have also been reported in human tracheobronchial epithelial cells. In this longitudinal study, we investigated systemic and local complement activation and NETosis in COVID-19 patients that underwent mechanical ventilation. Results confirmed significantly higher baseline levels of serum C5a (24.5 ± 39.0 ng/mL) and TCC (11.03 ± 8.52 µg/mL) in patients compared to healthy controls (p < 0.01 and p < 0.0001, respectively). Furthermore, systemic NETosis was significantly augmented in patients (5.87 (±3.71) × 106 neutrophils/mL) compared to healthy controls (0.82 (±0.74) × 106 neutrophils/mL) (p < 0.0001). In tracheal fluid, baseline TCC levels but not C5a and NETosis, were significantly higher in patients. Kinetic studies of systemic complement activation revealed markedly higher levels of TCC and CRP in nonsurvivors compared to survivors. In contrast, kinetic studies showed decreased local NETosis in tracheal fluid but comparable local complement activation in nonsurvivors compared to survivors. Systemic TCC and NETosis were significantly correlated with inflammation and coagulation markers. We propose that a ratio comprising systemic inflammation, complement activation, and chest X-ray score could be rendered as a predictive parameter of patient outcome in severe SARS-CoV-2 infections.


Subject(s)
COVID-19/immunology , Complement Activation/immunology , Inflammation/immunology , Aged , Aged, 80 and over , COVID-19/mortality , Complement C5a , Cytokines/blood , Epithelial Cells , Female , Humans , Inflammation/blood , Kinetics , Longitudinal Studies , Male , Prospective Studies , SARS-CoV-2 , Thorax/diagnostic imaging , Viral Load
14.
Future Microbiol ; 16: 1389-1400, 2021 12.
Article in English | MEDLINE | ID: covidwho-1528783

ABSTRACT

Background: We aimed to compare the clinical, laboratory and radiological findings of confirmed COVID-19 and unconfirmed patients. Methods: This was a single-center, retrospective study. Results: Overall, 620 patients (338 confirmed COVID-19 and 282 unconfirmed) were included. Confirmed COVID-19 patients had higher percentages of close contact with a confirmed or probable case. In univariate analysis, the presence of myalgia and dyspnea, decreased leukocyte, neutrophil and platelet counts were best predictors for SARS-CoV-2 RT-PCR positivity. Multivariate analyses revealed that only platelet count was an independent predictor for SARS-CoV-2 RT-PCR positivity. Conclusion: Routine complete blood count may be helpful for distinguishing COVID-19 from other respiratory illnesses at an early stage, while PCR testing is unique for the diagnosis of COVID-19.


Subject(s)
COVID-19/diagnosis , Adolescent , Adult , Aged , Aged, 80 and over , Blood Cell Count , COVID-19/blood , COVID-19/diagnostic imaging , COVID-19/virology , Female , Humans , Male , Middle Aged , Radiography , Retrospective Studies , SARS-CoV-2/classification , SARS-CoV-2/genetics , SARS-CoV-2/isolation & purification , Thorax/diagnostic imaging , Tomography, X-Ray Computed , Young Adult
15.
PLoS One ; 16(9): e0256630, 2021.
Article in English | MEDLINE | ID: covidwho-1518353

ABSTRACT

Pneumonia is a respiratory infection caused by bacteria or viruses; it affects many individuals, especially in developing and underdeveloped nations, where high levels of pollution, unhygienic living conditions, and overcrowding are relatively common, together with inadequate medical infrastructure. Pneumonia causes pleural effusion, a condition in which fluids fill the lung, causing respiratory difficulty. Early diagnosis of pneumonia is crucial to ensure curative treatment and increase survival rates. Chest X-ray imaging is the most frequently used method for diagnosing pneumonia. However, the examination of chest X-rays is a challenging task and is prone to subjective variability. In this study, we developed a computer-aided diagnosis system for automatic pneumonia detection using chest X-ray images. We employed deep transfer learning to handle the scarcity of available data and designed an ensemble of three convolutional neural network models: GoogLeNet, ResNet-18, and DenseNet-121. A weighted average ensemble technique was adopted, wherein the weights assigned to the base learners were determined using a novel approach. The scores of four standard evaluation metrics, precision, recall, f1-score, and the area under the curve, are fused to form the weight vector, which in studies in the literature was frequently set experimentally, a method that is prone to error. The proposed approach was evaluated on two publicly available pneumonia X-ray datasets, provided by Kermany et al. and the Radiological Society of North America (RSNA), respectively, using a five-fold cross-validation scheme. The proposed method achieved accuracy rates of 98.81% and 86.85% and sensitivity rates of 98.80% and 87.02% on the Kermany and RSNA datasets, respectively. The results were superior to those of state-of-the-art methods and our method performed better than the widely used ensemble techniques. Statistical analyses on the datasets using McNemar's and ANOVA tests showed the robustness of the approach. The codes for the proposed work are available at https://github.com/Rohit-Kundu/Ensemble-Pneumonia-Detection.


Subject(s)
COVID-19/diagnosis , Early Diagnosis , Pneumonia/diagnosis , Thorax/diagnostic imaging , COVID-19/diagnostic imaging , COVID-19/virology , Deep Learning , Humans , Lung/diagnostic imaging , Lung/pathology , Neural Networks, Computer , North America , Pneumonia/diagnostic imaging , SARS-CoV-2/isolation & purification , SARS-CoV-2/pathogenicity , Thorax/pathology , X-Rays
16.
Chest ; 160(2): 652-670, 2021 08.
Article in English | MEDLINE | ID: covidwho-1491830

ABSTRACT

The COVID-19 pandemic has had devastating medical and economic consequences globally. The severity of COVID-19 is related, in a large measure, to the extent of pulmonary involvement. The role of chest CT imaging in the management of patients with COVID-19 has evolved since the onset of the pandemic. Specifically, the description of CT scan findings, use of chest CT imaging in various acute and subacute settings, and its usefulness in predicting chronic disease have been defined better. We performed a review of published data on CT scans in patients with COVID-19. A summary of the range of imaging findings, from typical to less common abnormalities, is provided. Familiarity with these findings may facilitate the diagnosis and management of this disease. A comparison of sensitivity and specificity of chest CT imaging with reverse-transcriptase polymerase chain reaction testing highlights the potential role of CT imaging in difficult-to-diagnose cases of COVID-19. The usefulness of CT imaging to assess prognosis, to guide management, and to identify acute pulmonary complications associated with SARS-CoV-2 infection is highlighted. Beyond the acute stage, it is important for clinicians to recognize pulmonary parenchymal abnormalities, progressive fibrotic lung disease, and vascular changes that may be responsible for persistent respiratory symptoms. A large collection of multi-institutional images were included to elucidate the CT scan findings described.


Subject(s)
COVID-19/diagnostic imaging , Thorax/diagnostic imaging , Tomography, X-Ray Computed , COVID-19/complications , COVID-19/therapy , Humans , Prognosis , Sensitivity and Specificity
17.
Infect Dis Poverty ; 10(1): 126, 2021 Oct 21.
Article in English | MEDLINE | ID: covidwho-1477466

ABSTRACT

BACKGROUND: The computed tomography (CT) diagnostic value of COVID-19 is controversial. We summarized the value of chest CT in the diagnosis of COVID-19 through a meta-analysis based on the reference standard. METHODS: All Chinese and English studies related to the diagnostic value of CT for COVID-19 across multiple publication platforms, was searched for and collected. Studies quality evaluation and plotting the risk of bias were estimated. A heterogeneity test and meta-analysis, including plotting sensitivity (Sen), specificity (Spe) forest plots, pooled positive likelihood ratio (+LR), negative likelihood ratio (-LR), dignostic odds ratio (DOR) values and 95% confidence interval (CI), were estimated. If there was a threshold effect, summary receiver operating characteristic curves (SROC) was further plotted. Pooled area under the receiver operating characteristic curve (AUROC) and 95% CI were also calculated. RESULTS: Twenty diagnostic studies that represented a total of 9004 patients were included from 20 pieces of literatures after assessing all the aggregated studies. The reason for heterogeneity was caused by the threshold effect, so the AUROC = 0.91 (95% CI: 0.89-0.94) for chest CT of COVID-19. Pooled sensitivity, specificity, +LR, -LR from 20 studies were 0.91 (95% CI: 0.88-0.94), 0.71 (95% CI: 0.59-0.80), 3.1(95% CI: 2.2-4.4), 0.12 (95% CI: 0.09-0.17), separately. The I2 was 85.6% (P = 0.001) by Q-test. CONCLUSIONS: The results of this study showed that CT diagnosis of COVID-19 was close to the reference standard. The diagnostic value of chest CT may be further enhanced if there is a unified COVID-19 diagnostic standard. However, please pay attention to rational use of CT.


Subject(s)
COVID-19 , Thorax , Tomography, X-Ray Computed , COVID-19/diagnostic imaging , Humans , Reverse Transcriptase Polymerase Chain Reaction , Sensitivity and Specificity , Thorax/diagnostic imaging
18.
PLoS One ; 16(10): e0257884, 2021.
Article in English | MEDLINE | ID: covidwho-1468160

ABSTRACT

Recent studies show the potential of artificial intelligence (AI) as a screening tool to detect COVID-19 pneumonia based on chest x-ray (CXR) images. However, issues on the datasets and study designs from medical and technical perspectives, as well as questions on the vulnerability and robustness of AI algorithms have emerged. In this study, we address these issues with a more realistic development of AI-driven COVID-19 pneumonia detection models by generating our own data through a retrospective clinical study to augment the dataset aggregated from external sources. We optimized five deep learning architectures, implemented development strategies by manipulating data distribution to quantitatively compare study designs, and introduced several detection scenarios to evaluate the robustness and diagnostic performance of the models. At the current level of data availability, the performance of the detection model depends on the hyperparameter tuning and has less dependency on the quantity of data. InceptionV3 attained the highest performance in distinguishing pneumonia from normal CXR in two-class detection scenario with sensitivity (Sn), specificity (Sp), and positive predictive value (PPV) of 96%. The models attained higher general performance of 91-96% Sn, 94-98% Sp, and 90-96% PPV in three-class compared to four-class detection scenario. InceptionV3 has the highest general performance with accuracy, F1-score, and g-mean of 96% in the three-class detection scenario. For COVID-19 pneumonia detection, InceptionV3 attained the highest performance with 86% Sn, 99% Sp, and 91% PPV with an AUC of 0.99 in distinguishing pneumonia from normal CXR. Its capability of differentiating COVID-19 pneumonia from normal and non-COVID-19 pneumonia attained 0.98 AUC and a micro-average of 0.99 for other classes.


Subject(s)
Artificial Intelligence , COVID-19/diagnostic imaging , Pneumonia/diagnostic imaging , Thorax/diagnostic imaging , Humans , Predictive Value of Tests , Radiography, Thoracic , Sensitivity and Specificity
19.
Sci Rep ; 11(1): 19638, 2021 10 04.
Article in English | MEDLINE | ID: covidwho-1450291

ABSTRACT

The main purpose of this work is to investigate and compare several deep learning enhanced techniques applied to X-ray and CT-scan medical images for the detection of COVID-19. In this paper, we used four powerful pre-trained CNN models, VGG16, DenseNet121, ResNet50,and ResNet152, for the COVID-19 CT-scan binary classification task. The proposed Fast.AI ResNet framework was designed to find out the best architecture, pre-processing, and training parameters for the models largely automatically. The accuracy and F1-score were both above 96% in the diagnosis of COVID-19 using CT-scan images. In addition, we applied transfer learning techniques to overcome the insufficient data and to improve the training time. The binary and multi-class classification of X-ray images tasks were performed by utilizing enhanced VGG16 deep transfer learning architecture. High accuracy of 99% was achieved by enhanced VGG16 in the detection of X-ray images from COVID-19 and pneumonia. The accuracy and validity of the algorithms were assessed on X-ray and CT-scan well-known public datasets. The proposed methods have better results for COVID-19 diagnosis than other related in literature. In our opinion, our work can help virologists and radiologists to make a better and faster diagnosis in the struggle against the outbreak of COVID-19.


Subject(s)
COVID-19/diagnosis , Deep Learning , COVID-19/virology , Humans , Image Processing, Computer-Assisted , SARS-CoV-2/isolation & purification , Thorax/diagnostic imaging , Tomography, X-Ray Computed , X-Rays
20.
Medicine (Baltimore) ; 100(38): e22571, 2021 Sep 24.
Article in English | MEDLINE | ID: covidwho-1437852

ABSTRACT

BACKGROUND: There are few reports on the chest computed tomography (CT) imaging features of children with coronavirus disease 2019 (COVID-19), and most reports involve small sample sizes. OBJECTIVES: To systematically analyze the chest CT imaging features of children with COVID-19 and provide references for clinical practice. DATA SOURCES: We searched PubMed, Web of Science, and Embase; data published by Johns Hopkins University; and Chinese databases CNKI, Wanfang, and Chongqing Weipu. METHODS: Reports on chest CT imaging features of children with COVID-19 from January 1, 2020 to August 10, 2020, were analyzed retrospectively and a meta-analysis carried out using Stata12.0 software. RESULTS: Thirty-seven articles (1747 children) were included in this study. The heterogeneity of meta-analysis results ranged from 0% to 90.5%. The overall rate of abnormal lung CT findings was 63.2% (95% confidence interval [CI]: 55.8%-70.6%), with a rate of 61.0% (95% CI: 50.8%-71.2%) in China and 67.8% (95% CI: 57.1%-78.4%) in the rest of the world in the subgroup analysis. The incidence of ground-glass opacities was 39.5% (95% CI: 30.7%-48.3%), multiple lung lobe lesions was 65.1% (95% CI: 55.1%-67.9%), and bilateral lung lesions was 61.5% (95% CI: 58.8%-72.2%). Other imaging features included nodules (25.7%), patchy shadows (36.8%), halo sign (24.8%), consolidation (24.1%), air bronchogram signs (11.2%), cord-like shadows (9.7%), crazy-paving pattern (6.1%), and pleural effusion (9.1%). Two articles reported 3 cases of white lung, another reported 2 cases of pneumothorax, and another 1 case of bullae. CONCLUSIONS: The lung CT results of children with COVID-19 are usually normal or slightly atypical. The lung lesions of COVID-19 pediatric patients mostly involve both lungs or multiple lobes, and the common manifestations are patchy shadows, ground-glass opacities, consolidation, partial air bronchogram signs, nodules, and halo signs; white lung, pleural effusion, and paving stone signs are rare. Therefore, chest CT has limited value as a screening tool for children with COVID-19 and can only be used as an auxiliary assessment tool.


Subject(s)
COVID-19/diagnostic imaging , Lung/diagnostic imaging , Thorax/diagnostic imaging , Tomography, X-Ray Computed/statistics & numerical data , Adolescent , Blister/diagnostic imaging , Blister/epidemiology , Blister/virology , COVID-19/diagnosis , COVID-19/epidemiology , COVID-19/virology , Child , Child, Preschool , Data Management , Female , Humans , Incidence , Infant , Lung/pathology , Lung/virology , Male , Pleural Effusion/diagnostic imaging , Pleural Effusion/epidemiology , Pleural Effusion/virology , Pneumothorax/diagnostic imaging , Pneumothorax/epidemiology , Retrospective Studies , SARS-CoV-2/genetics , Solitary Pulmonary Nodule/diagnostic imaging , Solitary Pulmonary Nodule/epidemiology , Solitary Pulmonary Nodule/virology , Tomography, X-Ray Computed/methods , Tomography, X-Ray Computed/trends
SELECTION OF CITATIONS
SEARCH DETAIL