Citation

Chen, Yuanyuan and Khan, Waqas and Ali, Farman and Afsar, Haleem and Ejaz, Munaza and Alshamrani, Ali and Kamal, Shahid (2025) Cutting-edge bayesian deep learning and statistical strategies for bias mitigation in COVID-19 detection via chest x-ray imaging. Scientific Reports, 16 (1). ISSN 2045-2322

Text s41598-025-28723-x.pdf - Published Version Restricted to Repository staff only Download (4MB)

Abstract

Chest radiography (CXR is widely used for triage and follow-up of pulmonary disease, yet COVID-19 classification remains vulnerable to bias, label noise, and domain shift. We propose a multi-stage Bayesian deep learning framework that combines lung segmentation, segmentation-guided classification, calibrated ensembling, and uncertainty estimation to classify four classes (COVID-19, normal, viral pneumonia, bacterial pneumonia) and to grade COVID-19 severity. Models are trained and tested on 1,531 CXRs (100 COVID-19 images from 70 patients; 1,431 non-COVID images from ChestX-ray14) with patient-wise splits. The final ensemble achieves 98.33% test accuracy; COVID-19 sensitivity reaches 100% on this split. Robustness is quantified by stress-testing five image degradations (Gaussian noise, motion/defocus blur, JPEG compression, and downsampling), with macro AUC drops remaining small at moderate severities and larger under strong blur or heavy downsampling. Saliency and context-relevance analyses are used to identify spurious cues. The study is limited by dataset size and lack of external multi-site validation; a planned evaluation on COVIDx and BIMCV-COVID19+is outlined.

Item Type:	Article
Uncontrolled Keywords:	COVID-19 detection, Chest X-ray images, Baysian deep learning
Subjects:	Q Science > Q Science (General) > Q300-390 Cybernetics R Medicine > RA Public aspects of medicine > RA421-790.95 Public health. Hygiene. Preventive medicine
Divisions:	Faculty of Computing and Informatics (FCI)
Depositing User:	Ms Rosnani Abd Wahab
Date Deposited:	10 Feb 2026 02:06
Last Modified:	10 Feb 2026 02:06
URII:	http://shdl.mmu.edu.my/id/eprint/15265

Downloads

Edit (login required)