AigoHack

Inspiration

We wanted to apply AI in real life, helping clinicians in their goal and curating higher quality healthcare. Our proposed flow is meant to save clinicians time and effort, helping them focus on what needs their attention the most in order to give the best quality care possible. It was inspiring to see the promising result from the ACSE2019 competition. It encouraged us to integrate and implement the winning solution and further think about how it can be made a practical solution for real life scenarios.

What it does

Cobb angle estimation for screening of Scoliosis patients.
Annotation flow that captures expert annotation of cobb angles as well as building masks that help the model focus on the region of interest (spine). This annotation flow was incorporated to enable active learning through feedback from clinicians as well as to grow the dataset used for model development seamlessly.

How we built it

Through an iterative process between our team and the specialized mentors and clinicians, as we wanted our solution to be easy to use and practical for real life usage.
By making it easy for the experts to give feedback to the system and iteratively improve our core model as we acknowledge how it is crucial for helathcare solutions to be very accurate as well as transparent results.
By using tools that can be easily integrated with clinicians systems, as clinicians already have a system they’re using and we don’t want them to have to learn a new complicated system.

Challenges we ran into

Data Scarcity.
Limited time for experimentation.
Lack of domain experience.
Researching how our system can be integrated with existing systems.

Accomplishments that we're proud of

By working with great team, we managed to build this solution in a limited time.
We designed the system to be easily open for extension and improvement, as this is crucial for solutions in real life.
learned new technology and domain expertise.