AI Tool Capable of Classifying Brain Tumors within Hours
The team demonstrated that their deep learning model - DEPLOY - can accurately classify brain tumors into 10 major subtypes, the Xinhua reported.
The AI tool analyzes microscopic pictures of a patient's brain tissue to make its classification.
Danh-Tai Hoang, the co-lead of the project from the Biological Data Science Institute at ANU, said that the current gold standard for diagnosing and categorizing brain tumors - a process called DNA methylation-based profiling - can take several weeks and is not widely available.
By comparison, the AI tool can deliver results within hours.
"Remarkably, DEPLOY achieved an unprecedented accuracy of 95 percent," Hoang said in a media release.
"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."
DEPLOY was trained and validated on a dataset of approximately 4,000 patients from the United States and Europe.
Hoang said it could be used in the future to add to a pathologist's initial diagnosis or even prompt a re-evaluation when it finds disparities.
The researchers believe it could eventually be used to help classify other types of cancer.
4155/v
