author: Researchers have developed DEPLOY, an AI tool that can classify brain tumors into 10 major subtypes with 95% accuracy. This tool analyzes microscopic images of tumor tissue and provides a faster and more accessible alternative to DNA methylation-based profiling. DEPLOY may also be used to classify other cancers.
Important facts:
- DEPLOY can classify brain tumors with 95% accuracy.
- The AI tool analyzes microscopic images of tumor tissue.
- DEPLOY is a faster and more accessible alternative to DNA methylation-based profiling.
sauce: Australian National University
A new AI tool to classify brain tumors more quickly and accurately has been developed by researchers at the Australian National University (ANU).
According to Dr. Danh-Tai Hoang, accuracy in tumor diagnosis and classification is critical to effective patient treatment.
“The current gold standard for identifying different types of brain tumors is DNA methylation-based profiling,” Dr. Huang said.
“DNA methylation acts like a switch, controlling gene activity and controlling which genes are turned on or off.
“However, the time-consuming nature of these types of tests can be a major drawback, often lasting several weeks or more when patients need to make rapid treatment decisions.
“Also, these tests may not be available in nearly every hospital in the world.”
To address these challenges, ANU researchers, in collaboration with experts at the National Cancer Institute, have developed a method to predict DNA methylation and classify brain tumors into 10 major subtypes. I developed a certain DEPLOY.
DEPLOY utilizes microscopic photographs of patient tissues called histopathology images.
The model was trained and validated on a large dataset of approximately 4,000 patients from across the United States and Europe.
“Remarkably, DEPLOY achieved an unprecedented accuracy rate of 95 percent,” Dr. Hoang said.
“Furthermore, when given a subset of 309 samples that were particularly difficult to classify, DEPLOY was able to provide a more clinically relevant diagnosis than the pathologist originally provided.
“This points to the future potential of DEPLOY as a complementary tool, adding to the pathologist's initial diagnosis or even prompting re-evaluation in cases of discordance.”
Researchers believe DEPLOY could eventually be used to classify other types of cancer.
About this brain tumor and AI research news
author: jessica fagan
sauce: Australian National University
contact: Jessica Fagan – Australian Nation University
image: Image credited to Neuroscience News
Original research: Closed access.
“Prediction of tumor type based on DNA methylation from histopathology of central nervous system tumors using deep learning” by Danh-Tai Hoang et al. natural medicine
abstract
Prediction of tumor type based on DNA methylation from histopathology of central nervous system tumors using deep learning
Accurate diagnosis of different types of central nervous system (CNS) tumors is important for optimal treatment. DNA methylation profiles, which capture the methylation status of thousands of individual CpG sites, are a state-of-the-art, data-driven means to increase diagnostic accuracy, but are time-consuming and not widely available.
Here, to address these limitations, we developed Deep lEarning from histoPathoLOGy and methYlation (DEPLOY), a deep learning model that classifies CNS tumors into 10 major categories based on their histopathology.
DEPLOY integrates three different components. The first classifies CNS tumors directly from slide images (“direct model”), and the second first generates predictions of DNA methylation beta values that are then used for tumor classification (“indirect model”) . Classify tumor type directly from routinely available patient statistics.
First, we can see that DEPLOY accurately predicts beta values from histopathological images.
A 10-class model trained on an internal dataset of 1,796 patients was then used to predict tumor categories on three independent external test datasets containing 2,156 patients, and We achieved an overall accuracy of 95% and a balanced accuracy of 91%. with high confidence.
These results demonstrate the potential for future use of DEPLOY to assist pathologists in diagnosing CNS tumors in a clinically relevant short period of time.
