ABSTRACT
Since the first case of COVID-19 was reported in December 2019, many studies have been carried out on artificial intelligence for the rapid diagnosis of the disease to support health services. Therefore, in this study, we present a powerful approach to detect COVID-19 and COVID-19 findings from computed tomography images using pre-trained models using two different datasets. COVID-19, influenza A (H1N1) pneumonia, bacterial pneumonia and healthy lung image classes were used in the first dataset. Consolidation, crazy-paving pattern, ground-glass opacity, ground-glass opacity and consolidation, ground-glass opacity and nodule classes were used in the second dataset. The study consists of four steps. In the first two steps, distinctive features were extracted from the final layers of the pre-trained ShuffleNet, GoogLeNet and MobileNetV2 models trained with the datasets. In the next steps, the most relevant features were selected from the models using the Sine-Cosine optimization algorithm. Then, the hyperparameters of the Support Vector Machines were optimized with the Bayesian optimization algorithm and used to reclassify the feature subset that achieved the highest accuracy in the third step. The overall accuracy obtained for the first and second datasets is 99.46% and 99.82%, respectively. Finally, the performance of the results visualized with Occlusion Sensitivity Maps was compared with Gradient-weighted class activation mapping. The approach proposed in this paper outperformed other methods in detecting COVID-19 from multiclass viral pneumonia. Moreover, detecting the stages of COVID-19 in the lungs was an innovative and successful approach.
ABSTRACT
Covid-19 is a disease that affects the upper and lower respiratory tract and has fatal consequences in individuals. Early diagnosis of COVID-19 disease is important. Datasets used in this study were collected from hospitals in Istanbul. The first dataset consists of COVID-19, viral pneumonia, and bacterial pneumonia types. The second dataset consists of the following findings of COVID-19: ground glass opacity, ground glass opacity, and nodule, crazy paving pattern, consolidation, consolidation, and ground glass. The approach suggested in this paper is based on artificial intelligence. The proposed approach consists of three steps. As a first step, preprocessing was applied and, in this step, the Fourier Transform and Gradient-weighted Class Activation Mapping methods were applied to the input images together. In the second step, type-based activation sets were created with three different ResNet models before the Softmax method. In the third step, the best type-based activations were selected among the CNN models using the local interpretable model-agnostic explanations method and re-classified with the Softmax method. An overall accuracy success of 99.15% was achieved with the proposed approach in the dataset containing three types of image sets. In the dataset consisting of COVID-19 findings, an overall accuracy success of 99.62% was achieved with the recommended approach.
ABSTRACT
BACKGROUND: At present, the diagnosis of COVID-19 depends on real-time reverse transcriptase polymerase chain reaction (RT-PCT). On imaging, computed tomography (CT) manifestations resemble those seen in viral pneumonias, with multifocal ground-glass opacities and consolidation in a peripheral distribution being the most common findings. Although these findings lack specificity for COVID-19 diagnosis on imaging grounds, CT could be used to provide objective assessment about the extension of the lung opacities, which could be used as an imaging surrogate for disease burden. Chest CT scan may be helpful in early diagnosing of COVID-19. OBJECTIVE: The current study investigated the diagnostic accuracy and false-positive rate of chest CT in detecting COVID-19 pneumoniain a population with clinical suspicion using RT-PCR testing as reference standard. MATERIALS AND METHODS: In this prospective single centerstudy performed on 612 cases with clinical suspicion of COVID-19, all adult symptomatic ED patients had both a CT scan and a PCR upon arrival at the hospital. CT results were compared with PCR test (s) and diagnostic accuracy was calculated. RESULTS: Between February 15, 2020 to July 15, 2020, 612 symptomatic ED patients were included. In total, 78.5% of patients had a positive PCR and 82.8% a positive CT, resulting in a sensitivity of 94.2%, specificity 76.4%, likelihood ratio (LR) + 2.94 and (LR) - 0.18. The PPV was 76.7% and NPV 94.1%. The sensitivity of the CT tended to be higher (100.0%) in those with severe risk pneumonia than in patients with low/medium risk pneumonia (90.3%, P = 0.42). In patients with sepsis, sensitivity was significantly higher than in those without sepsis (99.5 vs. 63.5%, P < 0.001). The diagnostic ability of chest CT was found to be rather high with 92.1%, concordance rate between findings of CT and PCR. In 48 (7.8%) patients discordant findings between CT and PCR were observed. The positive predictive values (PPV) and accuracy of chest CT in diagnosing COVID-19 were higher in patients ≥60 years than that in patients <60 years (P = 0.001 and 0.004, respectively). The specificity and NPV of chest CT in diagnosing COVID-19 were greater for women than that for men (P = 0.007 and 0.03, respectively); and no difference existed for sensitivity, PPV and accuracy (P = 0.43, 0.69 and 0.31, respectively). In most cases, the CT scan was considered suspicious for COVID-19, while the PCR was negative (37/48, 70.8%). In the majority of these, the diagnosis at discharge was pulmonary infection (n = 26; 74.3%). The current study included repeated PCRs and explored discordant test results, which showed that in about 45.9% of patients with false-positive CT scans, other viral pathogens were detected. The false-positive rate of CT findings in the diagnosis of COVID-19 pneumonia was 7.2%. CONCLUSION: High diagnostic accuracy of chest CT findings with typical and relatively atypical CT manifestations of COVID-19 leads to a low rate of missed diagnosis. Normal chest CT can be found in RT-PCR positive COVID-19 cases, and typical CT manifestations can be found in RT-PCR negative cases. Therefore, a combination of both CT and RT-PCR for future follow-up, management and medical surveillance is recommended considering the false-positive results of chest CT in the diagnosis of COVID-19 pneumonia.
ABSTRACT
OBJECTIVE: Aimed to summarize the characteristics of chest CT imaging in Chinese hospitalized patients with Coronavirus Disease 2019 (COVID-19) to provide reliable evidence for further guiding clinical routine. METHODS: PubMed, Embase and Web of Science databases were searched to identify relevant articles involving the features of chest CT imaging in Chinese patients with COVID-19. All data were analyzed utilizing R i386 4.0.0 software. Random-effects models were employed to calculate pooled mean differences. RESULTS: 19 retrospective studies (1332 cases) were included. The results demonstrated that the combined proportion of ground-glass opacities (GGO) was 0.79 (95% CI 0.68, 0.89), consolidation was 0.34 (95% CI 0.23, 0.47); mixed GGO and consolidation was 0.46 (95% CI 0.37; 0.56); air bronchogram sign was 0.41 (95% CI 0.26; 0.55); crazy paving pattern was 0.32 (95% CI 0.17, 0.47); interlobular septal thickening was 0.55 (95% CI 0.42, 0.67); reticulation was 0.30 (95% CI 0.12, 0.48); bronchial wall thickening was 0.24 (95% CI 0.11, 0.40); vascular enlargement was 0.74 (95% CI 0.64, 0.86); subpleural linear opacity was 0.28 (95% CI 0.12, 0.48); intrathoracic lymph node enlargement was 0.03 (95% CI 0.00, 0.07); pleural effusions was 0.03 (95% CI 0.02, 0.06). The distribution in lung: the combined proportion of central was 0.05 (95% CI 0.01, 0.11); peripheral was 0.74 (95% CI 0.62, 0.84); peripheral involving central was 0.38 (95% CI 0.19, 0.75); diffuse was 0.19 (95% CI 0.06, 0.32); unifocal involvement was 0.09 (95% CI 0.05, 0.14); multifocal involvement was 0.57 (95% CI 0.48, 0.68); unilateral was 0.16 (95% CI 0.10, 0.23); bilateral was 0.83 (95% CI 0.78, 0.89); The combined proportion of lobes involved (> 2) was 0.70 (95% CI 0.61, 0.78); lobes involved (⦠2) was 0.35 (95% CI 0.26, 0.44). CONCLUSION: GGO, vascular enlargement, interlobular septal thickening more frequently occurred in patients with COVID-19, which distribution features were peripheral, bilateral, involved lobes > 2. Therefore, based on chest CT features of COVID-19 mentioned, it might be a promising means for identifying COVID-19.