HUMMUS

The recent AI boom inspired our team to apply Neural Networks to healthcare. There is a shortage of experience doctors in the market, especially cancer specialists, that is why we made an attempt at automizing one part of their job.

What it does

Our Neural Network model looks at CT scans of suspected lung cancer patients and determines if they have a tumor or not.

How we built it

Two of our teammates have experience with building Neural Networks. We used a random forest classifier as our AI model and fed it around 8000 images to train it on real-world cancer patient data.

Challenges we ran into

The brain cancer data was pretty hard to find. Even harder was organizing it and getting it ready to run into our AI model. This was the first time we had to do this: convert jpg images into data usable by our Neural Network.

Another challenge was developing a GUI or a friendly user interface. This was another thing we had to learn from scratch.

Accomplishments that we're proud of

We are proud that we manage to get a working AI that can predict if a patient is positive or negative for brain cancer to an accuracy of 97%. Of course, this does not replace the job of a doctor, rather it makes it easier and creates another layer of safety for patients. Our team is proud with our healthcare-oriented program and its cause.

What we learned

We learn to program an AI/Neural Network, we learned convert raw image data into a usable form for machine learning purposes and we learned to make a friendly user interface for people that want to use our product to analyze their CT scan for possible brain tumors.