A comparative analysis of the binary and multiclass classified chest X-ray images of pneumonia and COVID-19 with ML and DL models DOI Creative Commons

Madhumita Pal,

Ranjan K. Mohapatra, Ashish K. Sarangi

и другие.

Open Medicine, Год журнала: 2025, Номер 20(1)

Опубликована: Янв. 1, 2025

Abstract Background The highly infectious coronavirus disease 2019 (COVID-19) is caused by severe acute respiratory syndrome 2, the seventh coronavirus. It longest pandemic in recorded history worldwide. Many countries are still reporting COVID-19 cases even fifth year of its emergence. Objective performance various machine learning (ML) and deep (DL) models was studied for image-based classification lungs infected with COVID-19, pneumonia (viral bacterial), normal from chest X-rays (CXRs). Methods K -nearest neighbour logistics regression as two ML models, Visual Geometry Group-19, Vision transformer, ConvMixer three DL were included investigation to compare brevity detection cases. Results Among investigated returned best result terms accuracy, recall, precision, F 1-score area under curve both binary well multiclass classification. pre-trained model outperformed other four classifying. As per observations, there 97.1% accuracy + pneumonia-infected lungs, 98% 82% bacterial viral lungs. performed better than these tried on CXR image databases. Conclusion suggested network effectively detected different types using imagery. This could help medical sciences timely accurate diagnoses through bioimaging technology use high-end bioinformatics tools.

Язык: Английский

Fortifying defenses: Tactical safety protocols for COVID-19 sub-variant JN.1 in healthcare and laboratory settings DOI Creative Commons
Shazima Sheereen, Mohnish Zulfikar Manva, Shamama Sheereen

и другие.

Journal of Family Medicine and Primary Care, Год журнала: 2025, Номер 14(1), С. 78 - 84

Опубликована: Янв. 1, 2025

A BSTRACT Introduction: Primary care physicians are crucial in fighting COVID-19, especially with the emergence of new JN.1 sub-variant. Measures to Reduce Risk: Given your direct exposure infected patients, it is imperative establish a protocol for triaging patients respiratory symptoms and uphold minimum distance 2 meters between primary physicians. Patients suspected or diagnosed sub-variant should be advised wear surgical masks their protection others protection. must also use personal protective equipment (PPE) maintain strict hand hygiene practices when dealing these patients. Patient samples treated as high risk contamination, laboratory procedures meticulously evaluated potential hazards. PPE tailored procedure. Conclusion: To protect health well-being physicians, who play critical role addressing challenges, essential strictly adhere infection control measures.

Язык: Английский

Процитировано

0
A comparative analysis of the binary and multiclass classified chest X-ray images of pneumonia and COVID-19 with ML and DL models DOI Creative Commons

Madhumita Pal,

Ranjan K. Mohapatra, Ashish K. Sarangi

и другие.

Open Medicine, Год журнала: 2025, Номер 20(1)

Опубликована: Янв. 1, 2025

Abstract Background The highly infectious coronavirus disease 2019 (COVID-19) is caused by severe acute respiratory syndrome 2, the seventh coronavirus. It longest pandemic in recorded history worldwide. Many countries are still reporting COVID-19 cases even fifth year of its emergence. Objective performance various machine learning (ML) and deep (DL) models was studied for image-based classification lungs infected with COVID-19, pneumonia (viral bacterial), normal from chest X-rays (CXRs). Methods K -nearest neighbour logistics regression as two ML models, Visual Geometry Group-19, Vision transformer, ConvMixer three DL were included investigation to compare brevity detection cases. Results Among investigated returned best result terms accuracy, recall, precision, F 1-score area under curve both binary well multiclass classification. pre-trained model outperformed other four classifying. As per observations, there 97.1% accuracy + pneumonia-infected lungs, 98% 82% bacterial viral lungs. performed better than these tried on CXR image databases. Conclusion suggested network effectively detected different types using imagery. This could help medical sciences timely accurate diagnoses through bioimaging technology use high-end bioinformatics tools.

Язык: Английский

Процитировано

0