Inspiration
Our idea came from an unexpected conversation. One of our team members’ sisters is a doctor, and she was talking about how frustrating it is that even after more than three years, she still isn’t allowed to actually perform surgeries on her own. She mentioned how it’s hard to stay motivated when you’re constantly studying but don’t get to apply it in a real way. That made us curious: What if there was a way for students and early doctors to actually practice procedures safely before stepping into the Operating Room? Not just for learning, but also to build confidence and get rid of that initial fear. That’s where the idea of NeuroPark started.
What it does
NeuroPark is an AR-based training tool that lets users practice brain procedures, especially for hydrocephalus. You can interact with a 3D brain model and insert a virtual needle while the system gives real-time feedback on angle, depth, and distance. It also evaluates whether the attempt is safe or risky, helping users understand mistakes instantly.
How we built it
We trained an AI model on MRI scans of brains with hydrocephalus so it could learn the correct needle placement. Then we built an AR app using Unity to model 3d brain and needle and deployed it on a Meta Quest headset. The 3D brain model was then connected to the AI, which compares the user’s needle position to the ideal one and gives real-time feedback, including whether the attempt passes or fails.
Challenges we ran into
One of the biggest challenges was finding good medical datasets to train our AI. We also struggled with converting the AR interaction into usable data, like figuring out the exact position, angle, and depth of the needle. On top of that, understanding medical data as engineering students was not easy.
Accomplishments that we're proud of
We were able to make a working prototype despite limited data by utilizing MRI scans and training our AI based on data extraction. Additionally, for the needle placement, we were able to simplify the problem using basic math and trigonometry to calculate position, angle, and depth, which made the system much more manageable. Most importantly, our proudest achievement is that we created something that actually gives real-time feedback and has real-life use!
What we learned
During the hackathon, we were exposed to a lot of new technology and challenges, allowing us to learn along the way. One of the main things we learned was how to work with AR and AI together, and how to deal with situations where data is limited. We also learned how to break down a complex medical problem into something we could solve with engineering concepts.
What's next for NeuroPark
Due to time restrictions, we weren't able to add more features like we wanted; however, for future development, we want to make the system more personalized by using patient-specific brain scans instead of a generic model. We also want to improve the feedback system and expand it to cover more types of surgeries so it can be used more widely in medical training.
Log in or sign up for Devpost to join the conversation.