Study Demonstrates Potential of AI in Treating Brain Cancer

A tool powered by artificial intelligence has demonstrated potential in assisting doctors in combating aggressive brain tumors by identifying specific traits that can aid in surgical procedures.

The tool, called the Cryosection Histopathology Assessment and Review Machine, or CHARM, examines the images to quickly find the genetic profile of a tumor called a glioma. This process currently takes days to weeks, said Kun-Hsing Yu, senior author of the paper. the report was published on Friday in the journal Med. Surgeons use detailed diagnoses to guide them during surgery, Yu said, and being able to get them quickly could improve patient outcomes and save them from multiple surgeries.

Although glioma varies in severity, the aggressive form, called glioblastoma, can lead to death in less than six months if left untreated. According to the American Association of Neurological Surgeons, only 17 percent of people with glioblastoma survive the second year of diagnosis.

Surgeons use information about the genetic profile of a glioma tumor to decide how much tissue to remove from a patient’s brain and whether to implant discs coated with a cancer-fighting drug. However, obtaining this information currently requires time-consuming testing.

Yu and his team of researchers trained a machine learning algorithm to do the job by showing it images of samples collected during brain surgery and then comparing its work to the diagnoses of these patients. CHARM learned to compare or outperform other artificial intelligence systems in identifying a tumor’s genetic profile.

Although the tool is not as accurate as current genetic tests, the computer system can predict a tumor’s profile almost instantly. The quick analysis could allow doctors to proceed with the right kind of treatment without additional time to reschedule and perform another operation, Yu said.

By using its results in conjunction with other data, “the clinician can better make the right decision on the spot,” said Yu, an assistant professor of biomedical informatics at Harvard Medical School whose lab led the study.

CHARM can also distinguish malignant tumor cells from benign cells and identify tumor grade, which measures its aggressiveness. These are calls that human pathologists can make during surgery, but the tool can eliminate the need for a 10- to 15-minute wait or for a pathologist to be on standby during surgery.

Although the study showed promise, CHARM has yet to be tested in real life, the researchers said in a press release.

Yu’s team’s work is part of a broader effort to better diagnose and treat cancer with the help of artificial intelligence. An editorial published in the June issue of Lancet Oncology highlighted the ability of some systems to accurately identify people at high risk of pancreatic, lung and breast cancer.