Abstract

This project aims to develop an automated classification model for chest X-ray images to detect pulmonary diseases, with a focus on pneumonia. Using Deep Learning techniques, particularly Convolutional Neural Networks (CNNs), the model is trained to distinguish between normal and pneumonia-affected images. The solution includes a web application that allows users to upload an X-ray image and receive an instant diagnosis prediction. The system is designed to assist healthcare professionals by providing a fast and accurate diagnostic tool.

Methodology

Methodology

The workflow of the project includes the following stages:

• Data Preprocessing:

  • Images are resized (e.g., to 150x150 pixels) for uniformity.
  • Pixel values are normalized to accelerate training.

• CNN Model Architecture:

  • Two convolutional layers with 32 and 64 filters respectively, using 3x3 kernels and ReLU activation.
  • Two max pooling layers with 2x2 filters to reduce spatial dimensions.
  • A flatten layer followed by a dense layer with 128 neurons and ReLU activation.
  • A final output layer with 1 neuron and sigmoid activation for binary classification.

• Model Compilation and Training:

  • Loss function: Binary Crossentropy.
  • Optimizer: Adam.
  • Evaluation metrics: Accuracy, Precision, Recall, AUC.
  • Training performed on labeled datasets (Normal vs Pneumonia).

• Evaluation:

  • The dataset was split into training, validation, and testing sets.
  • Model performance was evaluated using key metrics on the test set.

• Deployment:

  • The model was deployed as a web application using Flask for the backend.
  • Bootstrap was used for a responsive and user-friendly interface.
  • Users can upload chest X-ray images and receive a real-time prediction (Normal or Pneumonia).

Results

The model demonstrated strong performance on the test set:

• Precision: 0.89
• Recall: 1.00
• Accuracy: 93.75%
• AUC (Area Under the Curve): 0.98

These results indicate that the model effectively detects pneumonia from chest X-ray images. The perfect recall score is especially important in a medical context, ensuring that all pneumonia cases are correctly identified.