Introduction

For the Brainiacs project, our team built a web application designed to serve as an interactive rat brain atlas. When working with neurological imaging, one of the biggest challenges is classification accuracy, especially when dealing with limited sample sizes. To solve this, we implemented Convolutional Neural Networks (CNNs) on the backend and integrated image augmentation techniques to improve how the model classifies different brain structures without overfitting.

Setup

We built the core application using Python and Flask to handle the heavy lifting and data processing, paired with a React frontend to keep the user interface responsive and easy to navigate. Because research teams often need to discuss specific regions of interest together in real time, we also integrated Socket.IO. This allowed us to add a live chat feature directly into the platform so users can collaborate, share notes, and discuss specific atlas coordinates instantly without needing to switch context or jump onto a separate communication tool.

Github Repository