Back to projects

Explainable COVID-19 Chest X-ray Classification

View resource →

  • TensorFlow
  • medical imaging
  • explainable AI
  • computer vision

What it is

A pipeline that classifies chest X-rays into four classes (COVID, Normal, Lung Opacity, Viral Pneumonia) and — crucially — explains why.

It builds both a custom CNN and a VGG16 model, then applies saliency / explainability methods (Grad-CAM, Integrated Gradients, and a combined Integrated-Grad-CAM) to visualize what the network attends to. It also includes adversarial-robustness utilities to probe how stable those explanations are.

Stack

TensorFlow/Keras · tf-explain (Grad-CAM, Integrated Gradients) · OpenCV · COVID-19 Radiography dataset

Links