Back to all papers

Trustworthy pneumonia detection in chest X-ray imaging through attention-guided deep learning.

November 14, 2025pubmed logopapers

Authors

Slimi H,Balti A,Abid S,Sayadi M

Affiliations (2)

  • Research Laboratory SIME, ENSIT, University of Tunis, Tunis, Tunisia. [email protected].
  • Research Laboratory SIME, ENSIT, University of Tunis, Tunis, Tunisia.

Abstract

Pneumonia remains a significant global health threat, especially among children, the elderly, and immunocompromised individuals. Chest X-ray (CXR) imaging is commonly used for diagnosis, but manual interpretation is prone to errors and variability. To address these challenges, we propose a novel attention-guided deep learning framework that combines spatial, temporal, and biologically inspired processing for robust and interpretable pneumonia detection. Our method integrates convolutional operations for spatial feature extraction, gated recurrent mechanisms to capture temporal dependencies, and spike-based neural processing to mimic biological efficiency and improve noise tolerance. The inclusion of an attention mechanism enhances the model's interpretability by identifying clinically relevant regions within the images. We evaluated the proposed method on a publicly available CXR dataset, achieving a high accuracy of 99.35%, along with strong precision, recall, and F1-score. Extensive experiments demonstrate the model's robustness to various types of image distortions, including Gaussian blur, salt-and-pepper noise, and speckle noise. These results confirm the effectiveness, reliability, and transparency of the proposed approach, making it a promising tool for clinical deployment, particularly in low-resource healthcare environments.

Topics

Deep LearningPneumoniaRadiography, ThoracicRadiographic Image Interpretation, Computer-AssistedJournal Article

Ready to Sharpen Your Edge?

Join hundreds of your 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.