A team of researchers at the University Medical Center Utrecht and the Princess Máxima Center for pediatric oncology in Amsterdam have leveraged artificial intelligence to help surgeons quickly identify the type of brain tumor a patient has during surgery, with identification available in a matter of hours rather than up to a week. The team’s work is published in a recent article published in Nature.
Brain and spinal cord tumors often require surgical procedures to treat and, indeed, to understand the type of tumor a patient has, including how aggressively the tumor is growing. Under normal circumstances, surgeons need to send off a sample of the tumor they collect during surgery and that sample needs to be reviewed by a pathologist at a molecular level to understand the nature of the tumor. This delay means a surgeon does not have the full picture when they are opening a patient up to treat a tumor.
Now, researchers have leveraged artificial intelligence and learning algorithms to speed up this diagnostic process. Specifically, researchers are leveraging a technology that is able to read analyze DNA samples in real time. With this technology, researchers trained the tool with a database of DNA “snapshots,” or samples, that enable to the tool to accurately identify the tumor type. This data base can learn from up to millions of these DNA samples.
The samples used to train the new technology can from a biobank reserve at the PrincessMáxima Center for pediatric oncology. This reserve is a collection of tissue samples collected from various brain tumors that were used to train the new technology.
One of the most beneficial aspects of this new technology is that analysis can occur directly in an operating room. Surgeons can collect samples, send for analysis, and get results in up to 90 minutes. With such a quick analysis, surgeons are able to course correct and apply more appropriate surgical strategies to address the type of tumor. Surgeons have already begun implementing this tool at the Princess Máxima Center
Sources: Science Daily; Nature
