The ensemble deep learning model for novel COVID-19 on CT images

Tao, Zhou; Lu, Huiling; Yang, Zaoli; Qiu, Shi; Bingqiang, Huo; Dong, Yong

doi:10.1016/j.asoc.2020.106885

Cited by 211 publications

(155 citation statements)

References 26 publications

Supporting

Mentioning

149

Contrasting

Unclassified

Order By: Relevance

“…[45] K-Nearest Neighbor, Naïve Bayes, Support Vector Machine (SVM), and Ensemble CT 306 COVID-19 (＋), 306 COVID-19 (−) Zhou et al. [46] Ensemble Deep Learning Model CT 500 COVID-19 (＋), 500 COVID-19 (−) Gupta et al. [47] Integrated Stacking InstaCovNet-19 model X-ray 361 COVID-19 (＋), 365 Normal (−), 362 Pneumonia Aslan et al.…”

Section: Related Workmentioning

confidence: 99%

A new approach for computer-aided detection of coronavirus (COVID-19) from CT and X-ray images using machine learning methods

Saygılı

2021

Applied Soft Computing

108

View full text Add to dashboard Cite

The COVID-19 outbreak has been causing a global health crisis since December 2019. Due to this virus declared by the World Health Organization as a pandemic, the health authorities of the countries are constantly trying to reduce the spread rate of the virus by emphasizing the rules of masks, social distance, and hygiene. COVID-19 is highly contagious and spreads rapidly globally and early detection is of paramount importance. Any technological tool that can provide rapid detection of COVID-19 infection with high accuracy can be very useful to medical professionals. The disease findings on COVID-19 images, such as computed tomography (CT) and X-rays, are similar to other lung infections, making it difficult for medical professionals to distinguish COVID-19. Therefore, computer-aided diagnostic solutions are being developed to facilitate the identification of positive COVID-19 cases. The method currently used as a gold standard in detecting the virus is the Reverse Transcription Polymerase Chain Reaction (RT-PCR) test. Due to the high false-negative rate of this test and the delays in the test results, alternative solutions are sought. This study was conducted to investigate the contribution of machine learning and image processing to the rapid and accurate detection of COVID-19 from two of the most widely used different medical imaging modes, chest X-ray and CT images. The main purpose of this study is to support early diagnosis and treatment to end the coronavirus epidemic as soon as possible. One of the primary aims of the study is to provide support to medical professionals who are most worn out and working under intense stress during COVID-19 through smart learning methods and image classification models. The proposed approach was applied to three different public COVID-19 data sets and consists of five basic steps: data set acquisition, pre-processing, feature extraction, dimension reduction, and classification stages. Each stage has its sub-operations. The proposed model performs in considerable levels of COVID-19 detection for dataset-1 (CT), dataset-2 (X-ray) and dataset-3 (CT) with the accuracy of 89.41%, 99.02%, 98.11%, respectively. On the other hand, in the X-ray data set, an accuracy of 85.96% was obtained for COVID-19 (+), COVID-19 (-), and those with Pneumonia but not COVID-19 classes. As a result of the study, it has been shown that COVID-19 can be detected with a high success rate in about less than one minute with image processing and classical learning methods. In the light of the findings, it is possible to say that the proposed system will help radiologists in their decisions, will be useful in the early diagnosis of the virus, and can distinguish pneumonia caused by the COVID-19 virus from the pneumonia of other diseases.

show abstract

Section: Related Workmentioning

confidence: 99%

A new approach for computer-aided detection of coronavirus (COVID-19) from CT and X-ray images using machine learning methods

Saygılı

2021

Applied Soft Computing

108

View full text Add to dashboard Cite

show abstract

“…Evaluation Indicator. In this paper, six evaluation indexes are selected to measure the experimental results: accuracy, sensitivity, specificity, Matthews correlation coefficient (MCC), F 1 score [13], and training time, and they are calculated by true positive (TP), false positive (FP), true negative (TN), and false negative (NN). Besides, TP indicates that the normal image is predicted to be normal, FP indicates that the abnormal image is predicted to be normal, TN indicates that lung tumor images are predicted as lung tumor images, and FN indicates that the normal lung image is predicted to be abnormal.…”

Section: Gradient Calculation Of the Convolutional Layermentioning

confidence: 99%

“…Zhou et al [ 12 ] propose a lung tumor Computer-aided diagnosis model in chest CT image based on DenseNet-NSCR (Non-negative, Sparse and collaborative representation classification of DenseNet) in this paper; the result shows that the DenseNet+NSCR model has better robustness and generalization capabilities compared with AlexNet+SVM, AlexNet+SRC, AlexNet+NSCR, GoogleNet+SVM, GoogleNet+SRC, GoogleNet+NSCR, DenseNet-201+SVM, and DenseNet-201+SRC. Zhou et al [ 13 ] use AlexNet, GoogleNet, and ResNet to realize the ensemble deep learning model for novel COVID-19 on CT images. A novel method for stock trend prediction uses a graph convolutional feature-based convolutional neural network (GC-CNN) model, in which both stock market information and individual stock information are considered in Chen et al [ 14 ].…”

Section: Introductionmentioning

confidence: 99%

Computer-Aided Diagnosis Research of a Lung Tumor Based on a Deep Convolutional Neural Network and Global Features

2021

BioMed Research International

Self Cite

View full text Add to dashboard Cite

Based on the better generalization ability and the feature learning ability of the deep convolutional neural network, it is very significant to use the DCNN on the computer-aided diagnosis of a lung tumor. Firstly, a deep convolutional neural network was constructed according to the fuzzy characteristics and the complexity of lung CT images. Secondly, the relation between model parameters (iterations, different resolution) and recognition rate is discussed. Thirdly, the effects of different model structures for the identification of a lung tumor were analyzed by changing convolution kernel size, feature dimension, and depth of the network. Fourthly, the different optimization methods on how to influence the DCNN performance were discussed from three aspects containing pooling methods (maximum pooling and mean pooling), activation function (sigmoid and ReLU), and training algorithm (batch gradient descent and gradient descent with momentum). Finally, the experimental results verified the feasibility of DCNN used on computer-aided diagnosis of lung tumors, and it can achieve a good recognition rate when selecting the appropriate model parameters and model structure and using the method of gradient descent with momentum.

show abstract

“…From a different perspective, CT scans include significantly more information compared to X-ray images. The interested reader can explore the recent work of Yang et al [23] or Zhou et al [24] for screening COVID-19 or for the prediction of the disease severity/ monitoring. The ImagingCOVID19AI European initiative, which gathers several hospitals from Europe, is of much interest, given the current technological difficulties, due to the lack of COVID-19 images.…”

Section: -Phase Chexpert-based Networkmentioning

confidence: 99%

Reliable Learning with PDE-Based CNNs and DenseNets for Detecting COVID-19, Pneumonia, and Tuberculosis from Chest X-Ray Images

Marginean

Muntean

et al. 2021

Mathematics

View full text Add to dashboard Cite

It has recently been shown that the interpretation by partial differential equations (PDEs) of a class of convolutional neural networks (CNNs) supports definition of architectures such as parabolic and hyperbolic networks. These networks have provable properties regarding the stability against the perturbations of the input features. Aiming for robustness, we tackle the problem of detecting changes in chest X-ray images that may be suggestive of COVID-19 with parabolic and hyperbolic CNNs and with domain-specific transfer learning. To this end, we compile public data on patients diagnosed with COVID-19, pneumonia, and tuberculosis, along with normal chest X-ray images. The negative impact of the small number of COVID-19 images is reduced by applying transfer learning in several ways. For the parabolic and hyperbolic networks, we pretrain the networks on normal and pneumonia images and further use the obtained weights as the initializers for the networks to discriminate between COVID-19, pneumonia, tuberculosis, and normal aspects. For DenseNets, we apply transfer learning twice. First, the ImageNet pretrained weights are used to train on the CheXpert dataset, which includes 14 common radiological observations (e.g., lung opacity, cardiomegaly, fracture, support devices). Then, the weights are used to initialize the network which detects COVID-19 and the three other classes. The resulting networks are compared in terms of how well they adapt to the small number of COVID-19 images. According to our quantitative and qualitative analysis, the resulting networks are more reliable compared to those obtained by direct training on the targeted dataset.

show abstract

The ensemble deep learning model for novel COVID-19 on CT images

Cited by 211 publications

References 26 publications

A new approach for computer-aided detection of coronavirus (COVID-19) from CT and X-ray images using machine learning methods

A new approach for computer-aided detection of coronavirus (COVID-19) from CT and X-ray images using machine learning methods

Computer-Aided Diagnosis Research of a Lung Tumor Based on a Deep Convolutional Neural Network and Global Features

Reliable Learning with PDE-Based CNNs and DenseNets for Detecting COVID-19, Pneumonia, and Tuberculosis from Chest X-Ray Images

Contact Info

Product

Resources

About