A Robust Framework for Epidemic Analysis, Prediction and Detection of COVID-19

Frontiers in Public Health, Journal Year: 2022, Volume and Issue: 10

Published: May 6, 2022

Covid-19 has become a pandemic that affects lots of individuals daily, worldwide, and, particularly, the widespread disruption in numerous countries, namely, US, Italy, India, Saudi Arabia. The timely detection this infectious disease is mandatory to prevent quick spread globally and locally. Moreover, COVID-19 coming time significant well cope with control by Governments. common symptoms COVID are fever as dry cough, which similar normal flu. devastating spreads quickly, all ages, aged people those feeble immune systems. There standard method employed detect COVID, real-time polymerase chain reaction (RT-PCR) test. But shortcomings, i.e., it takes long generates maximum false-positive cases. Consequently, we necessitate propose robust framework for estimation cases globally. To achieve above goals, proposed novel technique analyze, predict, infection. We made dependable estimates on parameters predictions infection potential washout frames countries used publicly available dataset composed Johns Hopkins Center estimation, analysis, during period 21 April 2020 27 June 2020. simple circulation fast model estimated Gaussian curve, utilizing parameter, least-square parameter curve fitting distinct areas. Forecasts depend upon results evolution central limit theorem data Covid prediction be justified. For gaussian distribution, parameters, extreme thickness regulated using statistical Y 2 fit aim doubling times after COVID-19, also technique, employing two features, Histogram Oriented Gradients Scale Invariant Feature Transform. designed CNN-based architecture named COVIDDetectorNet classification purposes. fed extracted features into viral pneumonia, other lung infections. Our obtained an accuracy 96.51, 92.62, 86.53% two, three, four classes, respectively. Experimental outcomes illustrate our reliable forecast disease.

Language: Английский

---

Classification of brain tumours from MR images with an enhanced deep learning approach using densely connected convolutional network

Computer Methods in Biomechanics and Biomedical Engineering Imaging & Visualization, Journal Year: 2022, Volume and Issue: 11(2), P. 266 - 277

Published: May 16, 2022

Brain cancer is one of the most leading causes death in human beings. There are different types tumours affecting brain and early diagnosis them increases survival rate. Classification from MR images an essential task treatment disease. Manual classification may lead to intra inter observer variability also time consuming. Hence, automated method assists doctors diagnosis, tumours. Since past decade, deep learning based methods widely used for problems especially medical image classification. In this paper, tumour proposed on enhanced approach using densely connected convolutional network (DenseNet). The transfer with DenseNet121 architecture CNN model optimised by tuning hyper-parameters network, thereby improving accuracy. evaluated publicly available data set comprising 3064 belonging three tumors – meningioma, glioma pituitary It inferred that DenseNet gives better accuracy compared VGG16, SVM AlexNet. Through hyper-parameter top dense layers CNN, improves 5.26% architecture. performs superior state-of-the-art a 97.39%.

Language: Английский

---

Classification of metastatic hepatic carcinoma and hepatocellular carcinoma lesions using contrast‐enhanced CT based on EI‐CNNet

Medical Physics, Journal Year: 2023, Volume and Issue: 50(9), P. 5630 - 5642

Published: March 4, 2023

Abstract Background For hepatocellular carcinoma and metastatic hepatic carcinoma, imaging is one of the main diagnostic methods. In clinical practice, diagnosis mainly relied on experienced physicians, which was inefficient cannot met demand for rapid accurate diagnosis. Therefore, how to efficiently accurately classify two types liver cancer based an urgent problem be solved at present. Purpose The purpose this study use deep learning classification model help radiologists single enhanced features CT (Computer Tomography) portal phase images site. Methods retrospective study, 52 patients with 50 were among who underwent preoperative examinations from 2017–2020. A total 565 slices these used train validate network (EI‐CNNet, training/validation: 452/113). First, EI block extract edge information enrich fine‐grained them. Then, ROC (Receiver Operating Characteristic) curve evaluate performance, accuracy, recall EI‐CNNet. Finally, results EI‐CNNet compared popular models. Results By utilizing 80% data training 20% validation, average accuracy experiment 98.2% ± 0.62 (mean standard deviation (SD)), rate 97.23% 2.77, precision 98.02% 2.07, parameters 11.83 MB, validation time 9.83 s/sample. improved by 20.98% base CNN 10.38 Compared other networks, InceptionV3 showed results, but number increased 33 s/sample, 6.51% using method. Conclusion demonstrated promised performance has potential reduce workload may distinguish whether tumor primary or in time; otherwise, it missed misjudged.

Language: Английский

---

A high-accuracy lightweight network model for X-ray image diagnosis: A case study of COVID detection

PLoS ONE, Journal Year: 2024, Volume and Issue: 19(6), P. e0303049 - e0303049

Published: June 18, 2024

The Coronavirus Disease 2019(COVID-19) has caused widespread and significant harm globally. In order to address the urgent demand for a rapid reliable diagnostic approach mitigate transmission, application of deep learning stands as viable solution. impracticality many existing models is attributed excessively large parameters, significantly limiting their utility. Additionally, classification accuracy model with few parameters falls short desirable levels. Motivated by this observation, present study employs lightweight network MobileNetV3 underlying architecture. This paper incorporates dense block capture intricate spatial information in images, well transition layer designed reduce size channel number feature map. Furthermore, label smoothing loss inter-class similarity effects uses class weighting tackle problem data imbalance. applies pruning technique eliminate unnecessary structures further parameters. As result, improved achieves an impressive 98.71% on openly accessible database, while utilizing only 5.94 million Compared previous method, maximum improvement reaches 5.41%. Moreover, research successfully reduces parameter count up 24 times, showcasing efficacy our approach. demonstrates benefits regions limited availability medical resources.

Language: Английский

---

DenseNet-121 Model for Diagnosis of COVID-19 Using Nearest Neighbour Interpolation and Adam Optimizer

Wireless Personal Communications, Journal Year: 2024, Volume and Issue: 137(3), P. 1823 - 1841

Published: July 8, 2024

Language: Английский

---