Related Experiment Videos
EXPLAIN COVIDNET: EXPLAINABLE AI FOR TRANSPARENT AND RELIABLE X-RAY SCREENING
Yogita Hande1, Ashwini V Zadgaonkar2, Rupali Vairagade3
1Department of Computer Engineering and Technology, Dr. Vishwanath Karad MIT World Peace University, Survey No, 124, Paud Rd, Kothrud, Pune, Maharashtra, 411038 India.
Seminars in Ultrasound, CT, and MR
|July 30, 2026
Summary
Explain-COVIDNet improves COVID-19 detection from X-rays using deep learning. The framework offers enhanced interpretability and accuracy, aiding clinical diagnosis.
Area of Science:
- Artificial Intelligence
- Medical Imaging
- Deep Learning
Background:
- Deep learning models show high accuracy but often lack interpretability and clinical reliability for medical image analysis.
- Current methods for COVID-19 detection from chest X-rays face challenges with noise sensitivity and transparency.
Purpose of the Study:
- To introduce Explain-COVIDNet, a novel framework for enhanced COVID-19 detection from chest X-ray images.
- To address limitations of existing deep learning models, focusing on interpretability, noise sensitivity, and clinical utility.
Main Methods:
- Preprocessing using Wavelet Contrast Enhancer (WCE) with Discrete Wavelet Transform (DWT) and Contrast-Limited Adaptive Histogram Equalization (CLAHE).
- Deep feature extraction and classification using a 154-layer Medical GoogLeNet architecture with Leaky ReLU activation.
- Incorporation of XGrad-CAM for model explainability, generating heatmaps of diagnostically relevant lung regions.
Main Results:
- Achieved high performance on benchmark datasets: up to 95.73% accuracy, 95.72% F1-score, 99.39 AUC, and 91.22 Cohen's Kappa.
- Demonstrated computational efficiency alongside strong diagnostic accuracy.
- Generated class-specific heatmaps highlighting key lung regions, enhancing model interpretability.
Conclusions:
- Explain-COVIDNet provides a reliable and interpretable deep learning solution for automated COVID-19 diagnosis using chest X-rays.
- The framework's enhanced preprocessing and explainability features contribute to its clinical relevance.
- Results indicate Explain-COVIDNet's potential as a valuable tool in medical diagnostics.
