Tuberculosis.AI

Inspiration

• Tuberculosis (TB) is a potentially serious infectious disease that mainly affects the lungs.
• A total of 1.5 million people died from TB in 2020 (WHO)
• Ending the TB epidemic by 2030 is among the health targets of the United Nations Sustainable Development Goals
• Tuberculosis is particularly difficult to diagnose in children.
• Early diagnosis can be a game changer and save millions of lives

What it does

Created a deep learning model to assist doctors to distinguish between Normal/Healthy Chests and chests with TB.

How we built it

• Dataset: Kaggle - https://www.kaggle.com/tawsifurrahman/tuberculosis-tb-chest-xray-dataset
  • 3500 Normal Chest X-rays | 700 Tuberculosis Chest X-rays

I took that dataset, cleaned, preprocessed and organized it. Created a custom neural network.

Challenges we ran into

• Time management & Project Management.
• Figuring out and designing a Deep learning model (via tensorflow)
• Computing power. Deep learning requires a good GPU to run smoothly, else it can take a while to get a model.

Accomplishments that we're proud of

• Successfully finishing the hackathon
• Learned few new things about deep learning
• Made something that has a positive impact
• High accuracy in the model (train and test)

What we learned

• Learned few new things about deep learning
• Learned how to manage time more efficiently

What's next for Tuberculosis.AI

• Mobile application and a web application I want to make it available for free so more people can be helped.

Built With

• deep-learning
• python
• tensorflow