the author: The researchers 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, providing a faster and more accessible alternative to DNA methylation-based profiling. DEPLOY can potentially be used to classify other cancers as well.
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.
Source: Australian National University
Researchers at the Australian National University (ANU) have developed a new AI tool to classify brain tumors more quickly and accurately.
According to Dr. Don Tai Huang, accuracy in tumor diagnosis and classification is crucial for effective patient treatment.
“The current gold standard for identifying different types of brain tumors is DNA methylation-based profiling,” said Dr. Huang.
“DNA methylation acts like a switch to control gene activity, and which genes are turned on or off.
“But the time it takes to do this type of testing can be a major drawback, often requiring several weeks or more when patients rely on quick decisions about treatment.
“Nearly all hospitals around the world also lack the availability of these tests.”
To address these challenges, ANU researchers, in collaboration with experts from the National Cancer Institute in the United States (US), developed DEPLOY, which predicts DNA methylation and subsequent brain tumors. There is a way of classifying into 10 major subtypes.
DEPLOY captures microscopic pictures of patient tissue called histopathology images.
The model was trained and validated on large datasets of nearly 4,000 patients from across the United States and Europe.
“Remarkably, DEPLOY achieved an unprecedented accuracy of 95 percent,” said Dr. Huang.
“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 that initially provided by the pathologist.
“This demonstrates the potential future role of DEPLOY as a complementary tool, augmenting the pathologist's initial diagnosis, or even prompting re-diagnosis in case of discrepancy.”
The researchers believe that DEPLOY could eventually be used to help classify other types of cancer as well.
About this brain cancer and AI research news
the author: Jessica Fagan
Source: Australian National University
contact: Jessica Fegan – Australian Nation University
Image: Photo credit to Neuroscience News.
Original research: closed access
“Prediction of tumor types based on DNA methylation from histopathology in central nervous system tumors with deep learning” by Don Tai Huang et al. Nature Medicine
Abstract
Prediction of tumor types based on DNA methylation from histopathology in central nervous system tumors with deep learning
Accuracy in the diagnosis of diverse central nervous system (CNS) tumor types is critical for optimal treatment. DNA methylation profiles, which capture the methylation status of thousands of individual CpG sites, are the most advanced data-driven means to increase diagnostic accuracy but are also time-consuming and not widely available.
Here, to overcome these limitations, we developed Deep Learning from Histopathology and Methylation (DEPLOY), a deep learning model that classified CNS tumors into ten major categories from histopathology. Is.
DEPLOY integrates three distinct components: the first classifies CNS tumors directly from slide images ('direct model'), the second initially generates predictions for DNA methylation beta values; , which are later used for tumor classification (the 'indirect model'), and a third classifies tumor types directly from a routinely available patient population.
First, we find that DEPLOY accurately predicts beta values from histopathology images.
Second, using a ten-class model trained on an internal dataset of 1,796 patients, we predict tumor categories with an overall accuracy of 95% and a balanced accuracy of 91% in three independent external test datasets including 2,156 patients. The predicted results are obtained. with high confidence.
These results demonstrate the potential future use of DEPLOY to aid pathologists in the diagnosis of CNS tumors within a clinically relevant short time frame.