Covid-19 Radiology

This project applies deep learning to detect COVID-19 from chest radiographs (X-ray images). It uses the COVID-19 Radiography Database from Kaggle, which contains chest X-ray images of patients with COVID-19, viral pneumonia, and normal (healthy) cases.

Goal

The goal of the project is to build a reliable image classification model that can distinguish COVID-19 cases from other conditions, achieving high performance with a balanced trade-off between precision and recall.

Approach

- Data Preparation

- Load and preprocess the COVID-19 Radiography Database.

- Resize and normalize images.

- Apply data augmentation (rotations, flips, etc.) to reduce overfitting and improve generalization.

Model Development

- Use VGG19, a pre-trained convolutional neural network, through transfer learning.

- Replace the top layers with a custom classification head suited to the dataset classes.

- Fine-tune the network on the radiology dataset.

Training

- Split data into training, validation, and test sets.

- Train with suitable hyperparameters (batch size, learning rate, epochs).

- Use callbacks such as early stopping and learning rate scheduling.

Evaluation

- Assess the model using metrics like accuracy, precision, recall, F1-score, and confusion matrix.

- Achieved an F1-score of 95%, showing strong and balanced classification performance.

Outcomes

- High performance: The project demonstrates that transfer learning with VGG19 is effective for COVID-19 radiology classification.

- Reproducibility: Uses a well-known dataset, enabling others to replicate and extend the results.

- Practical Relevance: Highlights the potential of deep learning to support medical diagnosis in radiology, though real-world deployment would require clinical validation.