RESUMO
In the early diagnosis of the Coronavirus disease (COVID-19), it is of great importance for either distinguishing severe cases from mild cases or predicting the conversion time that mild cases would possibly convert to severe cases. This study investigates both of them in a unified framework by exploring the problems such as slight appearance difference between mild cases and severe cases, the interpretability, the High Dimension and Low Sample Size (HDLSS) data, and the class imbalance. To this end, the proposed framework includes three steps: (1) feature extraction which first conducts the hierarchical segmentation on the chest Computed Tomography (CT) image data and then extracts multi-modality handcrafted features for each segment, aiming at capturing the slight appearance difference from different perspectives; (2) data augmentation which employs the over-sampling technique to augment the number of samples corresponding to the minority classes, aiming at investigating the class imbalance problem; and (3) joint construction of classification and regression by proposing a novel Multi-task Multi-modality Support Vector Machine (MM-SVM) method to solve the issue of the HDLSS data and achieve the interpretability. Experimental analysis on two synthetic and one real COVID-19 data set demonstrated that our proposed framework outperformed six state-of-the-art methods in terms of binary classification and regression performance.
RESUMO
We present COVID-CT-Mask-Net model that predicts COVID-19 in chest CT scans. The model works in two stages: in the first stage, Mask R-CNN is trained to localize and detect two types of lesions in images. In the second stage, these detections are fused to classify the whole input image. To develop the solution for the three-class problem (COVID-19, Common Pneumonia and Control), we used the COVIDx-CT data split derived from the dataset of chest CT scans collected by China National Center for Bioinformation. We use 3000 images (about 5% of the train split of COVIDx-CT) to train the model. Without any complicated data normalization, balancing and regularization, and training only a small fraction of the model's parameters, we achieve a 9 0 . 8 0 % COVID-19 sensitivity, 9 1 . 6 2 % Common Pneumonia sensitivity and 9 2 . 1 0 % true negative rate (Control sensitivity), an overall accuracy of 9 1 . 6 6 % and F1-score of 9 1 . 5 0 % on the test data split with 21192 images, bringing the ratio of test to train data to 7.06. We also establish an important result that regional predictions (bounding boxes with confidence scores) detected by Mask R-CNN can be used to classify whole images. The full source code, models and pretrained weights are available on https://github.com/AlexTS1980/COVID-CT-Mask-Net.
RESUMO
Whole cardiac segmentation in chest CT images is important to identify functional abnormalities that occur in cardiovascular diseases, such as coronary artery disease (CAD) detection. However, manual efforts are time-consuming and labor intensive. Additionally, labeling the ground truth for cardiac segmentation requires the extensive manual annotation of images by the radiologist. Due to the difficulty in obtaining the annotated data and the required expertise as an annotator, an unsupervised approach is proposed. In this paper, we introduce a semantic whole-heart segmentation combining K-Means clustering as a threshold criterion of the mean-thresholding method and mathematical morphology method as a threshold shifting enhancer. The experiment was conducted on 500 subjects in two cases: (1) 56 slices per volume containing full heart scans, and (2) 30 slices per volume containing about half of the top of heart scans before the liver appears. In both cases, the results showed an average silhouette score of the K-Means method of 0.4130. Additionally, the experiment on 56 slices per volume achieved an overall accuracy (OA) and mean intersection over union (mIoU) of 34.90% and 41.26%, respectively, while the performance for the first 30 slices per volume achieved an OA and mIoU of 55.10% and 71.46%, respectively.
Assuntos
Semântica , Tomografia Computadorizada por Raios X , Algoritmos , Análise por Conglomerados , Humanos , Processamento de Imagem Assistida por ComputadorRESUMO
The prevalence of the COVID-19 virus and its variants has influenced all aspects of our life, and therefore, the precise diagnosis of this disease is vital. If a polymerase chain reaction test for a subject is negative, but he/she cannot easily breathe, taking a computed tomography (CT) image from his/her lung is urgently recommended. This study aims to optimize a deep convolution neural network (DCNN) structure to increase the COVID-19 diagnosis accuracy in lung CT images. This paper employs the sine-cosine algorithm (SCA) to optimize the structure of DCNN to take raw CT images and determine their status. Three improvements based on regular SCA are proposed to enhance both the accuracy and speed of the results. First, a new encoding approach is proposed based on the internet protocol (IP) address. Then, an enfeebled layer is proposed to generate a variable-length DCNN. The suggested model is examined over the COVID-CT and SARS-CoV-2 datasets. The proposed method is compared to a standard DCNN and seven variable-length models in terms of five known metrics, including sensitivity, accuracy, specificity, F1-score, precision, and receiver operative curve (ROC) and precision-recall curves. The results demonstrate that the proposed DCNN-IPSCA surpasses other benchmarks, achieving final accuracy of (98.32% and 98.01%), the sensitivity of (97.22% and 96.23%), and specificity of (96.77% and 96.44%) on the SARS-CoV-2 and COVID-CT datasets, respectively. Also, the proposed DCNN-IPSCA performs much better than the standard DCNN, with GPU and CPU training times, which are 387.69 and 63.10 times faster, respectively.
Assuntos
COVID-19 , Algoritmos , COVID-19/diagnóstico por imagem , Teste para COVID-19 , Feminino , Humanos , Masculino , Redes Neurais de Computação , SARS-CoV-2 , Tomografia Computadorizada por Raios X/métodosRESUMO
This paper proposes an optimal structured deep convolutional neural network (DCNN) based on the marine predator algorithm (MPA) to construct a novel automatic diagnosis platform that may help radiologists identify COVID-19 and non-COVID-19 patients based on CT scan categorization and analysis. The goal is met with the help of three modifications based on the regular MPA. First, a novel encoding scheme based on Internet Protocol (IP) addresses is proposed, followed by introducing an Enfeebled layer to build a variable-length DCNN. Finally, the learning process divides big datasets into smaller chunks that are randomly evaluated. The proposed model is compared to the COVID-CT and SARS-CoV-2 datasets to undertake a complete evaluation. Following that, the performance of the developed model (DCNN-IPMPA) is compared to that of a typical DCNN and seven variable-length models using five well-known comparison metrics, as well as the receiver operating characteristic and precision-recall curves. The results show that the DCNN-IPMPA outperforms other benchmarks, with a final accuracy of 97.21% on the SARS-CoV-2 dataset and 97.94% on the COVID-CT dataset. Also, timing analysis indicates that the DCNN processing time is the best among all benchmarks as expected; however, DCNN-IPMPA represents a competitive result compared to the standard DCNN.
RESUMO
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.
Assuntos
COVID-19/diagnóstico por imagem , Tórax/diagnóstico por imagem , Tomografia Computadorizada por Raios X , COVID-19/complicações , COVID-19/terapia , Humanos , Prognóstico , Sensibilidade e EspecificidadeRESUMO
The novel COVID-19 is a global pandemic disease overgrowing worldwide. Computer-aided screening tools with greater sensitivity are imperative for disease diagnosis and prognosis as early as possible. It also can be a helpful tool in triage for testing and clinical supervision of COVID-19 patients. However, designing such an automated tool from non-invasive radiographic images is challenging as many manually annotated datasets are not publicly available yet, which is the essential core requirement of supervised learning schemes. This article proposes a 3D Convolutional Neural Network (CNN)-based classification approach considering both the inter-and intra-slice spatial voxel information. The proposed system is trained end-to-end on the 3D patches from the whole volumetric Computed Tomography (CT) images to enlarge the number of training samples, performing the ablation studies on patch size determination. We integrate progressive resizing, segmentation, augmentations, and class-rebalancing into our 3D network. The segmentation is a critical prerequisite step for COVID-19 diagnosis enabling the classifier to learn prominent lung features while excluding the outer lung regions of the CT scans. We evaluate all the extensive experiments on a publicly available dataset named MosMed, having binary- and multi-class chest CT image partitions. Our experimental results are very encouraging, yielding areas under the Receiver Operating Characteristics (ROC) curve of 0 . 914 ± 0 . 049 and 0 . 893 ± 0 . 035 for the binary- and multi-class tasks, respectively, applying 5-fold cross-validations. Our method's promising results delegate it as a favorable aiding tool for clinical practitioners and radiologists to assess COVID-19.
RESUMO
BACKGROUND: Computed tomography (CT) scans are increasingly available in clinical care globally. They enable a rapid and detailed assessment of tissue and organ involvement in disease processes that are relevant to diagnosis and management, particularly in the context of the COVID-19 pandemic. OBJECTIVE: The aim of this paper is to identify differences in the CT scan findings of patients who were COVID-19 positive (confirmed via nucleic acid testing) to patients who were confirmed COVID-19 negative. METHODS: A retrospective cohort study was proposed to compare patient clinical characteristics and CT scan findings in suspected COVID-19 cases. A multivariable logistic model with LASSO (least absolute shrinkage and selection operator) selection for variables was used to identify the good predictors from all available predictors. The area under the curve (AUC) with 95% CI was calculated for each of the selected predictors and the combined selected key predictors based on receiver operating characteristic curve analysis. RESULTS: A total of 94 (56%) patients were confirmed positive for COVID-19 from the suspected 167 patients. We found that elderly people were more likely to be infected with COVID-19. Among the 94 confirmed positive patients, 2 (2%) patients were admitted to an intensive care unit. No patients died during the study period. We found that the presence, distribution, and location of CT lesions were associated with the presence of COVID-19. White blood cell count, cough, and a travel history to Wuhan were also the top predictors for COVID-19. The overall AUC of these selected predictors is 0.97 (95% CI 0.93-1.00). CONCLUSIONS: Taken together with nucleic acid testing, we found that CT scans can allow for the rapid diagnosis of COVID-19. This study suggests that chest CT scans should be more broadly adopted along with nucleic acid testing in the initial assessment of suspected COVID-19 cases, especially for patients with nonspecific symptoms.
Assuntos
Técnicas de Laboratório Clínico/métodos , Tórax/diagnóstico por imagem , Tomografia Computadorizada por Raios X , Adolescente , Adulto , Teste para COVID-19 , Infecções por Coronavirus/diagnóstico , Feminino , Humanos , Masculino , Pessoa de Meia-Idade , Curva ROC , Reprodutibilidade dos Testes , Estudos Retrospectivos , Adulto JovemRESUMO
Lung cancer is the leading cause of US cancer-related deaths. Lung cancer screening with a low radiation dose chest computed tomography scan is now standard of care for a high-risk eligible population. It is imperative for clinicians and surgeons to evaluate the trade-offs of benefits and harms, including the identification of many benign lung nodules, overdiagnosis, and complications. Integration of smoking cessation interventions augments the clinical benefits of screening. Screening programs must develop strategies to manage screening-detected findings to minimize potential harms. Further research should focus on how to improve patient selection, minimize harms, and facilitate access to screening.