🫁 Pneumonia Detection from Chest X-Ray Images using Deep Learning

An end-to-end deep learning system for automated pneumonia detection from chest X-ray images using ResNet50 transfer learning and Grad-CAM explainability.

---

📌 Project Overview

Pneumonia is a serious respiratory infection that can be life-threatening if not detected early. This project builds an AI-based medical image classification system that can assist in identifying pneumonia from chest X-ray scans.

The system leverages transfer learning with a pre-trained ResNet50 model and incorporates Grad-CAM visual explanations to highlight lung regions influencing the model’s decision.

---

🧠 Key Features

• Automated pneumonia detection from X-ray images
• Medical image classification using deep learning
• Transfer learning with ResNet50
• Fine-tuning for improved performance
• Full model evaluation with medical metrics
• Grad-CAM visual explanations for interpretability

---

🗂 Dataset

Source: Kaggle Chest X-Ray Pneumonia Dataset
Providers: NIH Clinical Center & Guangzhou Women and Children's Medical Center

The dataset contains chest X-ray images categorized into:

• Normal
• Pneumonia

⚠️ Dataset is not included in this repository due to size and licensing.

---

🏗 Model Architecture

• Base Model: ResNet50 (ImageNet pretrained)
• Input Size: 224 × 224
• Transfer Learning with frozen base layers
• Fine-tuning of top convolutional layers
• Output Layer: Binary classification (Normal vs Pneumonia)

---

⚙️ Technologies Used

• Python
• TensorFlow / Keras
• NumPy, Matplotlib, Seaborn
• OpenCV
• Google Colab

---

📊 Model Evaluation

The model was evaluated on an unseen test set using:

• Accuracy
• Precision
• Recall (Sensitivity)
• Specificity
• Confusion Matrix
• ROC Curve (AUC)

Medical emphasis was placed on high recall for pneumonia cases to reduce false negatives.

---

📊 Confusion Matrix

📈 ROC Curve

🔬 Model Explainability (Grad-CAM)

Grad-CAM was used to visualize which lung regions influenced the model’s predictions.

🧪 Pneumonia Case

🧪 Normal Case

These heatmaps show the model focuses on clinically relevant lung areas, increasing trust and transparency.

---

📁 Repository Structure

notebooks/
│
├── 01_Model_Training_Pneumonia.ipynb
├── 01B_FineTuning_Continuation.ipynb
├── 02_Model_Evaluation.ipynb
└── 03_GradCAM_Explainability.ipynb

gradcam_results/
├── gradcam_pneumonia.png
└── gradcam_normal.png

---

▶️ How to Run

1. Download the dataset from Kaggle

2. Place the dataset in Google Drive or your local path

3. Run the notebooks in this order:

   • 01_Model_Training_Pneumonia.ipynb
   • 01B_FineTuning_Continuation.ipynb
   • 02_Model_Evaluation.ipynb
   • 03_GradCAM_Explainability.ipynb

---

🎯 Applications

• AI-assisted pneumonia screening
• Medical image analysis research
• Healthcare AI systems
• Clinical decision-support tools

---

🚀 Future Improvements

• Train on larger, more diverse datasets
• Multi-class classification (viral vs bacterial pneumonia)
• Deployment as a web app for clinical use
• Ensemble models for improved robustness

---

⭐ If you found this project interesting, consider giving it a star!