Identifying the genetic characteristics of a glioma tumor is a process that traditionally takes days or weeks, but a study published July 7 in the journal Med revealed an artificial intelligence tool can predict a tumor's profile almost instantly.
The tool, the Cryosection Histopathology Assessment and Review Machine, or CHARM, is a machine-learning algorithm that was trained by researchers showing it sample photos collected during brain surgeries and comparing its work with each respective diagnosis, according to a July 7 report from Bloomberg.
Knowing the genetic profile of a tumor is necessary for surgeons to know how much brain tissue to remove during surgery and whether to implant drug-coated wafers to start fighting the cancer. The speed at which the tool can provide this information allows clinicians to make immediate decisions, eliminating time-consuming testing and helping patients avoid multiple surgeries, according to Kun-Hsing Yu, MD, PhD, assistant professor of biomedical informatics at Boston-based Harvard Medical School and author of the study.
CHARM was also able to distinguish malignant from benign tumors and grade a tumor's severity, the report said.
While the study revealed the tool is more effective than any other AI system, it is still not as accurate as current genetic tests. Researchers said it still needs to be tested in real-world settings before it can officially help physicians make treatment decisions, according to Bloomberg.