Pneumonia Detection Using CNN + LSTM Hybrid Model

CNN + LSTM Hybrid Model

Comment

Pneumonia Detection Using CNN + LSTM Hybrid Model

This project implements a hybrid deep learning model to detect pneumonia from pediatric chest X-ray images. By combining Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks, the model captures both local spatial features and global contextual dependencies across lung regions. Built using TensorFlow/Keras, it includes visualizations, performance metrics, and a user-friendly interface for real-world deployment. Dataset

We use the Pediatric Chest X-ray Dataset from Kaggle, which includes labeled chest X-ray images categorized as:

  • NORMAL
  • PNEUMONIA

Folder Structure:

Pediatric Chest X-ray Pneumonia/
├── train/
│   ├── NORMAL/
│   └── PNEUMONIA/
├── test/
│   ├── NORMAL/
│   └── PNEUMONIA/

Deep Learning Architecture Convolutional Neural Networks (CNNs)

CNNs extract spatial features such as edges, textures, and shapes from X-ray images.

Key Components:

  • Conv2D: Learns visual patterns
  • MaxPooling2D: Reduces dimensionality
  • ReLU: Adds non-linearity
  • He Initialization: Stabilizes gradient flow

Why CNN? CNNs are ideal for medical imaging tasks due to their ability to learn hierarchical spatial features. Long Short-Term Memory (LSTM)

LSTMs capture long-range dependencies across reshaped image sequences.

Key Components:

  • Memory Cell: Retains historical context
  • Gates: Control information flow
  • LSTM Layer: Processes feature sequences

Why LSTM? By reshaping CNN outputs into sequences, LSTMs can learn spatial relationships across lung regions—critical for detecting pneumonia spread. CNN + LSTM Hybrid Model

Architecture Flow:

Input Image (224x224x3)
↓
CNN Layers → Feature Maps
↓
Reshape → Sequence Format
↓
LSTM Layer → Contextual Understanding
↓
Dense Layers → Classification
↓
Softmax → Output: NORMAL or PNEUMONIA

Advantages:

  • CNN: Captures local lung features
  • LSTM: Understands global spatial dependencies

Model Compilation & Training

Training Details:

  • Loss Function: sparse_categorical_crossentropy
  • Optimizer: Adam
  • Metric: accuracy
  • Normalization: Pixel values scaled to [0, 1]
  • Epochs: 20
  • Validation Split: 20% Evaluation Metrics

Metrics Used:

  • Accuracy: Overall prediction correctness
  • Confusion Matrix: TP, FP, TN, FN breakdown
  • Classification Report: Precision, Recall, F1-score

Visualization

Tools Used:

  • Matplotlib: For plotting training curves
  • Seaborn: For heatmaps and confusion matrix

Grad-CAM Integration:

  • Highlights lung regions influencing predictions
  • Builds trust and interpretability for clinicians Summary of Key Concepts
Concept Purpose
CNN Extract spatial features from X-ray images
LSTM Understand spatial dependencies across image regions
Reshape Layer Convert CNN output to sequence format for LSTM
Dense + Softmax Final classification into NORMAL/PNEUMONIA
Image Normalization Scale image pixels to [0,1] for stable training
Model Evaluation Accuracy, confusion matrix, precision, recall
Visualization Helps interpret model performance visually
Grad-CAM Visual explanation of model predictions

The app lets users export a detailed PDF report containing patient metadata, prediction results, and visual Grad-CAM overlays. This feature supports clinical documentation and easy sharing with healthcare professionals or patients. With one click, users can securely download the report for offline review or integration into medical records.

Built With

Share this project:

Updates