Tuberculosis (TB) remains a critical global health challenge, requiring efficient and accurate diagnostic methods. This project focuses on leveraging machine learning techniques to classify chest X-ray images as either normal or tuberculosis-infected.
The dataset used consists of:
- Training Set: 2,940 chest X-ray images
- Testing Set: 1,260 chest X-ray images
This makes it a binary classification problem, where the goal is to correctly differentiate between normal and TB-positive chest X-rays.
The performance of models is assessed using:
- Accuracy
- Precision
- Recall
- F1-score (used as the final benchmark metric)
The classification model is built using deep learning techniques, ensuring robustness and generalization across unseen test data. The final model submission is evaluated against a predefined benchmark using the F1-score metric.
- Clone this repository:
git clone https://github.com/your-username/tb-detection.git cd tb-detection - Install dependencies:
pip install -r requirements.txt
- Run the model training:
python train.py
- Evaluate the model:
python evaluate.py
This project is licensed under the MIT License.