![](https://ainews.one/wp-content/uploads/2024/05/12A57EB5-8DC3-47B8-A1F3-2B74BCE906D2-802x628.jpg)
Researchers in Australia and the United States say that a new artificial intelligence tool has allowed them to classify brain tumors more quickly and accurately.
The current method for identifying different kinds of brain tumors, while accurate, can take several weeks to produce results. The method, called DNA methylation-based profiling, is not available at many hospitals around the world.
To address these challenges, a research team from the Australian National University, in collaboration with the National Cancer Institute in the United States, has developed a way to predict DNA methylation, which acts like a switch to control gene activity.
This allows them to classify brain tumors into 10 major categories using a deep learning model.
This is a branch of artificial intelligence that teaches computers to process data in a way that is inspired by a human brain.
The joint U.S.-Australian system is called DEPLOY and uses microscopic pictures of a patient’s tissue called histopathology images.
The researchers see the DEPLOY technology as complementary to an initial diagnosis by a pathologist or physician.
Danh-Tai Hoang, a research fellow at the Australian National University, told VOA that AI will enhance current diagnostic methods that can often be slow.
“The technique is very time consuming,” Hoang said. “It is often around two to three weeks to obtain a result from the test, whereas patients with high-grade brain tumors often require treatment as soon as possible because time is the goal for brain tumor(s), so they need to get treatment as soon as possible.”
The research team said its AI model was validated on large datasets of approximately 4,000 patients from across the United States and Europe and an accuracy rate of 95 percent.
Their study has been published in the journal Nature Medicine.