AI platform detects neurodegenerative disease in tissue samples

A convolutional neural network is able to identify neurofibrillary tangles, a primary marker of Alzheimer’s disease, with a high degree of accuracy directly from digitized images.

In fact, the artificial intelligence platform—developed by researchers at the Center for Computational and Systems Pathology at Mount Sinai—can detect a range of neurodegenerative disease in human brain tissue samples, including chronic traumatic encephalopathy.

While the accurate diagnosis of these diseases is challenging and requires highly trained specialists, a study conducted at Mount Sinai’s Icahn School of Medicine demonstrates that applying powerful machine learning approaches to digitized microscopic slides can recognize and quantify the buildup of abnormal tau proteins in the brain in neurofibrillary tangles (NFT) that occur in neurodegenerative diseases.

digital pathology.jpg

“We applied deep learning to the neuropathological assessment of NFT in postmortem human brain tissue to develop a classifier capable of recognizing and quantifying tau burden,” report the authors in their study published in the journal Laboratory Investigation. “We found that the deep learning framework is capable of identifying and quantifying NFT with a range of staining intensities and diverse morphologies.”

According to the researchers, this is the first framework available for evaluating deep learning algorithms using large-scale image data in neuropathology.

“Recently, powerful machine learning-based approaches have emerged, allowing the recognition and quantification of pathological changes from digital images,” states the study, which was funded with grants from the Department of Defense, the National Institutes of Health, Alzheimer’s Association and the Rainwater Charitable Foundation.

Also See: New tech could change how pathologists view tissue samples

“Utilizing artificial intelligence has great potential to improve our ability to detect and quantify neurodegenerative diseases, representing a major advance over existing labor-intensive and poorly reproducible approaches,” said lead investigator John Crary, MD, professor of pathology and neuroscience at the Icahn School of Medicine. “Ultimately, this project will lead to more efficient and accurate diagnosis of neurodegenerative diseases.”

Researchers from the Boston University School of Medicine, VA Boston Healthcare System and UT Southwestern Medical Center also contributed to this work.

For reprint and licensing requests for this article, click here.