Brain cancer can be difficult to diagnose because identifying key genetic markers typically requires surgery, which may delay treatment and can miss important tumor information. This project aims to develop an AI based “virtual biopsy” capable of predicting these molecular markers directly from MRI, eliminating the need for tissue samples. Using AI techniques such as self supervised learning, graph based models, and generative AI, the system will be trained on large datasets and adapted to patients at the Moncton Hospital. This approach will help address gaps in imaging data, maintain patient privacy, and ensure the model works reliably across different healthcare settings. Ultimately, the goal is to deliver a fast, accurate, and non invasive tool that supports earlier and more precise treatment planning for people with brain cancer.
