CXGNet: A tri-phase chest X-ray image classification for COVID-19 diagnosis using deep CNN with enhanced grey-wolf optimizer

Biomedical Signal Processing and Control, Journal Year: 2022, Volume and Issue: 77, P. 103860 - 103860

Published: June 6, 2022

Language: Английский

---

Inverted bell-curve-based ensemble of deep learning models for detection of COVID-19 from chest X-rays

Neural Computing and Applications, Journal Year: 2022, Volume and Issue: 35(22), P. 16113 - 16127

Published: Jan. 5, 2022

Abstract Novel Coronavirus 2019 disease or COVID-19 is a viral caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). The use of chest X-rays (CXRs) has become an important practice to assist in the diagnosis as they can be used detect abnormalities developed infected patients’ lungs. With fast spread disease, many researchers across world are striving several deep learning-based systems identify from such CXR images. To this end, we propose inverted bell-curve-based ensemble learning models for detection We first selection pretrained on ImageNet dataset and concept transfer retrain them with datasets. Then trained combined proposed bell curve weighted method, where output each classifier assigned weight, final prediction done performing average those outputs. evaluate method two publicly available datasets: Radiography Database IEEE COVID Chest X-ray Dataset. accuracy, F1 score AUC ROC achieved 99.66%, 99.75% 99.99%, respectively, dataset, and, 99.84%, 99.81% other dataset. Experimental results ensure that their combination using result improved predictions CXRs.

Language: Английский

---

Fuzzy rank-based fusion of CNN models using Gompertz function for screening COVID-19 CT-scans

Scientific Reports, Journal Year: 2021, Volume and Issue: 11(1)

Published: July 8, 2021

Abstract COVID-19 has crippled the world’s healthcare systems, setting back economy and taking lives of several people. Although potential vaccines are being tested supplied around world, it will take a long time to reach every human being, more so with new variants virus emerging, enforcing lockdown-like situation on parts world. Thus, there is dire need for early accurate detection prevent spread disease, even more. The current gold-standard RT-PCR test only 71% sensitive laborious perform, leading incapability conducting population-wide screening. To this end, in paper, we propose an automated system that uses CT-scan images lungs classifying same into COVID Non-COVID cases. proposed method applies ensemble strategy generates fuzzy ranks base classification models using Gompertz function fuses decision scores adaptively make final predictions Three transfer learning-based convolutional neural network used, namely VGG-11, Wide ResNet-50-2, Inception v3, generate be fused by model. framework been evaluated two publicly available chest CT scan datasets achieving state-of-the-art performance, justifying reliability relevant source codes related present work in: GitHub.

Language: Английский

---

ET-NET: an ensemble of transfer learning models for prediction of COVID-19 infection through chest CT-scan images

Multimedia Tools and Applications, Journal Year: 2021, Volume and Issue: 81(1), P. 31 - 50

Published: Aug. 31, 2021

Language: Английский

---

Transfer learning based novel ensemble classifier for COVID-19 detection from chest CT-scans

Computers in Biology and Medicine, Journal Year: 2021, Volume and Issue: 141, P. 105127 - 105127

Published: Dec. 11, 2021

Language: Английский

---

COVID-19 detection from CT scans using a two-stage framework

Expert Systems with Applications, Journal Year: 2022, Volume and Issue: 193, P. 116377 - 116377

Published: Jan. 1, 2022

Language: Английский

---

A hybrid CNN–KNN approach for identification of COVID-19 with 5-fold cross validation

Sensors International, Journal Year: 2023, Volume and Issue: 4, P. 100229 - 100229

Published: Jan. 1, 2023

The novel coronavirus is the new member of SARS family, which can cause mild to severe infection in lungs and other vital organs like heart, kidney liver. For detecting COVID-19 from images, traditional ANN be employed. This method begins by extracting features then feeding into a suitable classifier. classification rate not so high as feature extraction dependent on experimenters' expertise. To solve this drawback, hybrid CNN-KNN-based model with 5-fold cross-validation proposed classify covid-19 or non-covid19 CT scans patients. At first, some pre-processing steps contrast enhancement, median filtering, data augmentation, image resizing are performed. Secondly, entire dataset divided five equal sections folds for training testing. By doing cross-validation, generalization ensured overfitting network prevented. CNN consists four convolutional layers, max-pooling two fully connected layers combined 23 layers. architecture used extractor case. taken model's fourth layer finally, classified using K Nearest Neighbor rather than softmax better accuracy. conducted over an augmented 4085 scan images. average accuracy, precision, recall F1 score after performing 98.26%, 99.42%,97.2% 98.19%, respectively. method's accuracy comparable existing works described further, where state art custom models were used. Hence, diagnose patients higher efficiency.

Language: Английский

---

Hybrid Meta-Heuristic based Feature Selection Mechanism for Cyber-Attack Detection in IoT-enabled Networks

Procedia Computer Science, Journal Year: 2023, Volume and Issue: 218, P. 318 - 327

Published: Jan. 1, 2023

Today's technologically advanced connected world is mostly reliant on the Internet of Things (IoT)-enabled smart gadgets and easy connectivity. These are more susceptible to malicious practices found in network traffic, which one biggest challenges cyber security domain. As a result, many systems end-users adversely affected by this practice. However, intrusion detection (IDS) often applied guard against cyber-attacks. Since, IDS plays key role detecting preventing cyber-attacks IoT-enabled networks, but design an efficient fast system for cyber-attack still challenging research issue. Moreover, datasets contain multiple features IDS, feature selection (FS) essential mechanism remove irrelevant redundant from large datasets. Thus, paper has proposed hybrid scheme statistical test-based filter approaches such as Chi-Square (χ^2), Pearson's Correlation Coefficient (PCC), Mutual Information (MI) combined with Non-Dominated Sorting Genetic Algorithm (NSGA-II)-based metaheuristic approach optimization features. In scheme, filter-based methods employed rank guided population initialization NSGA-II faster convergence towards solution. Performance evaluation evaluated using ToN-IoT dataset terms number selected accuracy. Experimental outcomes compared some latest state-of-art techniques. Result analysis confirms superior performance minimum optimized (only 13 out 43 features) maximum accuracy (99.48%).

Language: Английский

---

Automated semantic lung segmentation in chest CT images using deep neural network

Neural Computing and Applications, Journal Year: 2023, Volume and Issue: 35(21), P. 15343 - 15364

Published: April 10, 2023

Language: Английский

---

Microstructural segmentation using a union of attention guided U-Net models with different color transformed images

Scientific Reports, Journal Year: 2023, Volume and Issue: 13(1)

Published: April 7, 2023

Metallographic images or often called the microstructures contain important information about metals, such as strength, toughness, ductility, corrosion resistance, which are used to choose proper materials for various engineering applications. Thus by understanding microstructures, one can determine behaviour of a component made particular metal, and predict failure that in certain conditions. Image segmentation is powerful technique determination morphological features microstructure like volume fraction, inclusion morphology, void, crystal orientations. These some key factors determining physical properties metal. Therefore, automatic micro-structure characterization using image processing useful industrial applications currently adopts deep learning-based models. In this paper, we propose metallographic method an ensemble modified U-Nets. Three U-Net models having same architecture separately fed with color transformed imaged (RGB, HSV YUV). We improvise dilated convolutions attention mechanisms get finer grained features. Then apply sum-rule-based on outcomes final prediction mask. achieve mean intersection over union (IoU) score 0.677 publicly available standard dataset, namely MetalDAM. also show proposed obtains results comparable state-of-the-art methods fewer number model parameters. The source code work be found at https://github.com/mb16biswas/attention-unet .

Language: Английский

---