If you’re researching artificial intelligence (AI) for cancer care, see how NCI funding is helping drive a real-life application in detecting and surgically removing brain tumors. Researchers are testing FastGlioma, an open-source foundational model that helps guide brain tumor surgery.

After tumor surgery, remains of malignant tissue can regrow, causing tumor recurrence and decreased patient survival. The rates of tumors left behind after surgery have not improved in the past two decades. Like you, the researchers who developed FastGlioma want to do something to improve those rates!

Here are some highlights from the FastGlioma research to consider if you’re interested in creating similar models for cancer care:

• FastGlioma uses simulated Raman histology for rapid tissue imaging that doesn’t rely on time consuming dyes or labels. This makes the tool more practical for your clinical user!
• FastGlioma is a general-purpose model for all diffuse gliomas and degrees of tumor infiltration.
• FastGlioma accesses real-time, accurate diagnostic information that surgeons can actually use.
• FastGlioma outperformed standard-of-care image-guided and fluorescence-guided methods for tumor detection by a large margin in simulated trials.

FastGlioma’s success shows the advantage of visual foundation models for medical AI applications. The results also highlight the potential for using visual foundation models to help you generalize to other cancers without the need for extensive model retraining.