• Posted November 13, 2024

AI Helps Spot Brain Tumor Tissue Surgeons Miss

A newly developed AI program can help doctors detect and potentially remove brain cancer that might otherwise be missed during surgery, a new study demonstrates.

The AI, called FastGlioma, calculated how much residual brain cancer remained following surgery with approximately 92% accuracy, researchers reported Nov. 13 in the journal Nature.

FastGlioma missed high-risk residual tumor just under 4% of the time, compared with a nearly 25% miss rate for human doctors relying on MRI scans or fluorescent dyes to detect brain tumors, results showed.

Not only that, but the AI can return these results within 10 seconds, making it a potentially powerful aid to surgeons in the middle of removing a brain tumor, researchers said.

“FastGlioma is an artificial intelligence-based diagnostic system that has the potential to change the field of neurosurgery by immediately improving comprehensive management of patients with diffuse gliomas,” said senior researcher Dr. Todd Hollon, a neurosurgeon at University of Michigan Health.

“The technology works faster and more accurately than current standard-of-care methods for tumor detection and could be generalized to other pediatric and adult brain tumor diagnoses,” Hollon added in a university news release. “It could serve as a foundational model for guiding brain tumor surgery.”  

Neurosurgeons are rarely able to remove the entire mass of a life-threatening brain tumor, leaving behind what doctors call residual tumor, researchers said in background notes.

This leftover brain cancer is missed because it often resembles healthy brain tissue along the edges of the cavity left behind by the removal of a tumor, researchers said.

Residual tumor increases the risk of a person’s cancer recurring, robs them of years of life and frequently requires follow-up brain surgeries, researchers said.

Doctors try to suss out residual tumor during surgery using MRI scans and fluorescent dyes that highlight tumor cells, but these technologies have limited usefulness, researchers said.

To improve surgeons’ ability to completely remove a brain tumor, researchers combined artificial intelligence with microscopic optical imaging to create FastGlioma.

The team trained FastGlioma’s AI using more than 11,000 surgical specimens and 4 million unique microscopic snapshots of healthy brain tissue and cancerous tumors.

Researchers then analyzed fresh, unprocessed specimens samples from 220 patients who had operations for brain cancers.

FastGlioma detected residual tumor with up to 92% accuracy within about 100 seconds using high-resolution images, and with 90% accuracy when using a “fast mode” that relies on slightly lower resolution images.

“This means that we can detect tumor infiltration in seconds with extremely high accuracy, which could inform surgeons if more resection is needed during an operation,” Hollon explained.

AI also can be taught to distinguish other cancer types from healthy tissue, researchers added.

“In future studies, we will focus on applying the FastGlioma workflow to other cancers, including lung, prostate, breast, and head and neck cancers,” said researcher Dr. Aditya Pandey, chair of neurosurgery at University of Michigan Health.

More information

The American Cancer Society has more about brain cancer.

SOURCE: University of Michigan, news release, Nov. 13, 2024