New AI Tool Classifies Brain Tumors More Quickly and Accurately

Precision in diagnosing and categorizing tumors is essential for delivering effective treatment to patients. Currently, the gold standard for identifying various types of brain tumors involves DNA methylation-based profiling. DNA methylation functions as a regulatory mechanism to control gene activity, essentially turning genes on or off. However, the time required for such testing can be a significant hindrance, often taking several weeks—a delay that can be critical when patients need prompt decisions regarding their treatment. Additionally, these tests are not widely available in most hospitals around the world. Now, a new artificial intelligence (AI) tool has been developed to classify brain tumors more quickly and accurately.

Researchers at The Australian National University (ANU, Canberra, Australia) have created DEPLOY. This new method predicts DNA methylation patterns to classify brain tumors into 10 major subtypes. DEPLOY utilizes histopathology images, which are microscopic images of patient tissue samples. The model was trained and validated on extensive datasets comprising around 4,000 patients from the US and Europe. DEPLOY demonstrated remarkable accuracy, achieving a rate of 95%. Additionally, in a subset of 309 samples that were particularly challenging to classify, DEPLOY provided diagnoses that were more clinically relevant than those initially given by pathologists. The researchers anticipate that DEPLOY could also be adapted for classifying other types of cancer in the future.

“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,” said Dr. Danh-Tai Hoang.

Related Links:
ANU

Visit expo >

Gold Member

TORCH Panel Rapid Test

Rapid TORCH Panel Test

Antipsychotic TDM Assays
Saladax Antipsychotic Assays

New

Gold Member

ANCA IFA

Kallestad Autoimmune ANCA IFA Complete Kit

New

Nanoparticle Analyzer

Videodrop