SCOLIOSIS HACKATHON
The AI session that provided a break through for Scoliosis sufferers
How it Works
The Idea
Detecting scoliosis in medical imagery using machine learning and image processing specifically, calculating the COBB angles from coronal radiographs.
How It Works
We calculated the angle between the upper points of one vertebra and the lower points of all the other vertebrae; we only kept the largest angle.
Conclusion
Our Solution was able to diagnose the COBB angle from radiograph images of scoliosis patients, with the ability to be introduced in almost orthopedic clinic.
Our Hackathon Plan
Modify our annotation app (Tis@) to work well with radiograph images
Create a neural network in PyTorch trained on the provided data to get the landmarks for all the upper and lower points of thoracolumbar vertebrae
- PyTorch
- Python
- Microsoft Azure