AI Reveals Hidden Brain Lesions in Multiple Sclerosis MRI
Despite decades of research, Multiple Sclerosis still holds many unanswered questions. MRI scans are regularly a crucial component for monitoring the disease progression of individuals suffering from Multiple Sclerosis. But there is one problem with traditional imaging. Its failure to capture any damage to the gray matter in the brain.
Doctors treating Multiple Sclerosis have faced one frustrating challenge for many years. Many patients continued to experience worsening memory, difficulty concentrating, and increasing disability, though the MRI scans looked stable. It was clear that something was happening inside the brain, and that could not be revealed by the traditional imaging.
But now, researchers are about to take an important step toward solving the mystery. In a new study conducted by a group of researchers from the University at Buffalo, a unique AI-based MRI technique that can reveal hidden Brain Lesions that cannot be seen with conventional MRI scans has been discovered. This new finding can aid in understanding the progression of Multiple Sclerosis.
The findings were recently published in the journal Communications Medicine.
Why do Hidden Brain Lesions Matter in Multiple Sclerosis?
Multiple Sclerosis is an autoimmune disease in which the immune system mistakenly attacks the body’s own protective covering around nerve fibres. This damage disrupts communication between the brain and the rest of the body. This could lead to symptoms such as vision problems, muscle weakness, balance issues, fatigue, and cognitive decline.
The MRI scan is usually preferred by doctors to diagnose the disease by detecting lesions in the white matter of the brain. White matter lesions are among the major indicators used in the diagnosis of Multiple Sclerosis.
Nevertheless, scientists are aware that the disease also affects the brain’s gray matter, which includes the cerebral cortex. These Brain Lesions, known as cortical lesions, are highly associated with issues with memory and cognition. However, they have largely gone unnoticed on MRIs until now.
Even though patients report worsening symptoms, this limitation has made it difficult for doctors to understand the disease progression completely.
AI in MRI Finds What Traditional Scans Could Miss!
Dr. Michael G. Dwyer, associate professor of neurology and biomedical informatics at UB, explained that the AI does not create new information but uncovers details that already exist within MRI scans. An AI in an MRI system has been created by the research team that can detect such hidden brain lesions through MRI images taken during clinical trials.
This innovative technique does not require any new imaging tools; rather makes use of advanced image processing and deep learning methods on already available MRI images.
The researchers stated that the algorithm does not work with single images from MRI scans. The algorithm compares various MRI contrasts at once and finds small distinctions that are hardly noticed by humans.
These small details allow the algorithm to form information that is invisible in traditional MRI scans, thus making Brain Lesions visible.
A New Technique Called MMCLE
To address this challenge, researchers have created a new imaging technique called the Multimodal Cortical Lesion Enhancement (MMCLE). It uses sophisticated image processing and deep learning to find cortical lesions undetectable by regular MRI scans. This approach was tested using MRI data collected during the Phase III ORATORIO clinical trial for 700+ Multiple Sclerosis patients treated with ocrelizumab.
More Than 11,000 Hidden Brain Lesions Detected
Dr. Dwyer noted that researchers have long known these cortical lesions existed because they could be observed in postmortem brain tissue. The challenge was finding a reliable way to detect them in living patients using routine MRI scans. Advances in deep learning have now made that possible.
Scientists have detected 15-20 previously invisible Brain Lesions per patient using AI in the MRI Scan. This reveals more than 11,000 cortical lesions across the entire dataset. The study also showed that these lesions could be measured consistently. Also, this could make the method valuable for future research and disease monitoring.
Why does it matter?
Traditional MRI results are often insufficient to explain why patients suffering from Multiple Sclerosis keep on having problems with memory or cognitive dysfunction. The fact that artificial intelligence in MRI allows the identification of hidden brain lesions can provide a full picture of the disease development for the researchers.
Unlocking Existing MRI Data
A major advantage of this method is its ability to work on already existing MRI images. This implies that thousands of MRI images used in past studies about multiple sclerosis can be analyzed again by using this technique.
The Future of AI in MRI for Multiple Sclerosis
Dr. Zivadinov believes the findings could help researchers revisit MRI data from past Multiple Sclerosis clinical trials, revealing disease patterns that were previously impossible to measure. The team also expects the technology to improve future studies by providing a more complete picture of disease progression.
Artificial intelligence in MRI is only in the research phase at present. It can potentially aid neurologists in monitoring Multiple Sclerosis with precision, predicting the progression of the disease, and assessing the efficacy of treatment in the future. Far from replacing the radiologist’s role, the technology complements MRI by identifying Brain Lesions that a regular MRI scan fails to detect.
