Back to all papers

EXPLAIN COVIDNET: EXPLAINABLE AI FOR TRANSPARENT AND RELIABLE X-RAY SCREENING.

July 30, 2026pubmed logopapers

Authors

Hande Y,Zadgaonkar AV,Vairagade R,Gunjal A

Affiliations (4)

  • Department of Computer Engineering and Technology, Dr. Vishwanath Karad MIT World Peace University, Survey No, 124, Paud Rd, Kothrud, Pune, Maharashtra, 411038 India. Electronic address: [email protected].
  • Department of Computer Engineering and Technology (DCET), Dr. Vishwanath Karad MIT World Peace University, Kothrud, Pune, Maharashtra, 411038 India.
  • Department of Cyber Security, Shah and Anchor Kutchhi Engineering College, Mahavir Trust Education, SAKEC, Mumbai, 400 088 India.
  • Department of Computer Engineering and Technology, Dr. Vishwanath Karad MIT World Peace University, Survey No, 124, Paud Rd, Kothrud, Pune, Maharashtra 411038 India.

Abstract

Recently, deep learning has emerged as a prominent branch of AI, gaining attention for its high accuracy and wide-ranging application in various domains. The proposed Explain-COVIDNet framework enhances COVID-19 detection from chest X-ray images by addressing key limitations in current deep learning models, including poor interpretability, sensitivity to noise, and limited clinical reliability. It begins with a robust preprocessing stage using a Wavelet Contrast Enhancer (WCE), which integrates Discrete Wavelet Transform (DWT) for denoising and Contrast-Limited Adaptive Histogram Equalization (CLAHE) for contrast enhancement. This results in high-quality input images that aid model accuracy. Medical GoogLeNet, a 154-layer architecture with inception modules, is used for deep feature extraction and classification, employing Leaky ReLU to retain subtle but crucial image details. To improve transparency, XGrad-CAM is incorporated for explainability, generating class-specific heatmaps that highlight diagnostically relevant lung regions. The model demonstrates strong performance on two benchmark datasets, achieving up to 95.73% accuracy, 95.72% F1-score, 99.39 AUC, and 91.22 Cohen's Kappa, while maintaining computational efficiency. These results underscore Explain-COVIDNet's potential as a reliable, interpretable, and clinically meaningful tool for automated COVID-19 diagnosis.

Topics

Journal Article

Ready to Sharpen Your Edge?

Subscribe to join 11k+ peers who rely on RadAI Slice. Get the essential weekly briefing that empowers you to navigate the future of radiology.

We respect your privacy. Unsubscribe at any time.