A new AI tool is helping doctors predict if brain cancer might come back in children. This could improve care for young patients with a type of brain tumor called glioma. These tumors can often be treated with surgery, but if the cancer comes back, it can be serious.
Researchers from Mass General Brigham, Boston Children’s Hospital, and Dana-Farber/Boston Children’s Cancer Center used AI to study brain scans of children after treatment. The goal was to spot early signs that the cancer might return. They trained a deep learning model using almost 4,000 MRI scans from 715 children.
Usually, AI models look at one scan at a time. But this new tool uses a method called “temporal learning,” where it studies several scans taken over months. This helps the AI see changes in the brain over time, which can signal cancer is coming back.
The results were promising. The AI predicted cancer relapse with 75% to 89% accuracy — much better than the 50% accuracy of models that only use one scan. The researchers found that looking at four to six scans gave the best results.
More testing is needed before this tool is used in real hospitals. But the hope is that it could reduce how often children need scans if they are low-risk or help doctors treat high-risk patients sooner.
“This is a big step in using AI to help children with brain tumors,” said Dr. Benjamin Kann, one of the lead researchers. “It could help doctors make better choices and reduce stress for families.”
Source: technologynetworks
