Researcher's AI tool will help save cancer patients' valuable time: ANU

The tool developed by a team, led by researcher Dr Danh-Tai Hoang can classify brain tumours within hours, a process that usually takes weeks depending on the hospital.

After training AI using histopathology images (microscopic pictures of tissue) from about 4000 patients from the US and Europe, Dr Hoang said the model returned results with "unprecedented" accuracy.

This would mean patients with tumours relying on such a classification for treatment options could have a better chance of beating the cancer.

"Brain tumours can be difficult to diagnose and the patient's time is valuable. This tool can help them get treatment as soon as possible," he said.

Dr Hoang said the present "gold standard" for detection was was at least 2-3 weeks to classify a subtype using DNA methylation-based profiling.

"DNA methylation acts like a switch to control gene activity, and which genes are turned on or off," he said.

Dr Hoang added a disadvantage of this method was DNA methylation tests were not available in all hospitals.

The AI tool, named DEPLOY, is the answer to this challenge and can "predict" brain tumour subtypes with 95 per cent accuracy.

DEPLOY was developed at ANU in collaboration with experts from the National Cancer Institute in the US.

"Furthermore, when given a subset of 309 particularly difficult to classify samples, DEPLOY was able to provide a diagnosis that was more clinically relevant than what was initially provided by pathologists," Dr Hoang said.

"This shows the potential future role of DEPLOY as a complementary tool, adding to a pathologist's initial diagnosis, or even prompting re-evaluation in the case of disparities."

The DEPLOY team believe the tool can eventually also be used to help classify other types of cancer. Their research has been published in peer-reviewed journal Nature Medicine.