Highly sensitive imaging technique detects myelin damage

(a) Widefield 4× tile scan of a coronal brain slice from a rhesus monkey with an induced cortical lesion (yellow arrow). qBRM RGB parameter maps are rendered where the brightness of each pixel maps the retardance (myelin density), and the optic-axis orientation (related to fiber direction) is mapped to the color wheel (a—top right). It should be noted that the optic axis of myelin lipids is radial around axons and thus orthogonal to the fiber direction (please refer to Blanke and Gray et al.) and is still meaningful for myelin debris, when there is no longer any axonal structure. (b) High-resolution 40× (NA 0.75) qBRM image of the perilesional gray matter reveals a clear accumulation of myelin debris (yellow arrows). (c) High-resolution 40× (NA 0.75) qBRM image of the corpus callosum, where myelin appears more intact compared to the perilesional region, but still exhibits signs of myelin breakdown (yellow arrows). (d) By contrast, the corpus callosum of a healthy, aging rhesus monkey from the same anatomical region shows little indication of myelin breakdown, demonstrating the specificity of injury-induced myelin pathology visualized by qBRM.

The breakdown of myelin, the insulating layer around brain cells that supports brain function, is prevalent in a range of neurodegenerative diseases, aging and because of various forms of trauma. While electron microscopy is considered the gold standard for ultrastructural imaging of myelin, it is considered impractical for large-scale studies due to its limited field of view and time-consuming and complex sample preparation requirements.

In a new study from Boston University Chobanian & Avedisian School of Medicine and BU’s College of Engineering, researchers used a special microscope called birefringence microscopy (BRM) paired with an automated deep learning algorithm to reliably count and map myelin damage across whole sections of the brain—something not feasible with other techniques. The ability to image and measure damage to myelin will lead to better understanding of the patterns and extent that occurs with disease, injury and normal aging.

The study is published in the journal Neurophotonics.

“A major advantage of BRM over conventional imaging methods is its ability to rapidly image large areas at high resolution without special staining, making it uniquely suited for studying widespread myelin pathology,” says corresponding author Alex Gray, Ph.D., ’25.

The researchers used two groups of experimental models that had sustained limited damage in the motor area of the brain, mimicking a stroke. Models treated with stem cell derived extracellular vesicles (a therapeutic treatment) fully recovered from injury, which was shown when the imaged brain sections were seen with the BRM.

Induction of cortical lesion to the primary motor cortex M1 and section preparation. Credit: Neurophotonics (2025). DOI: 10.1117/1.nph.12.4.045006

The researchers then trained an AI deep learning algorithm to automatically identify and quantify myelin damage across the brain. Lastly, they compared the amount and location of damage between treated and untreated models to relate tissue changes to recovery of function.

According to the researchers, this approach not only provides insights into the spatial distribution of myelin debris in this model, but also offers a framework for studying other models of myelin damage, ultimately contributing to a deeper understanding of the relationships between myelin structural integrity and functional and cognitive deficits.

“This can guide development and testing of therapies that protect or restore neural wiring. It may help research into stroke and ischemic injury, chronic traumatic encephalopathy (CTE), multiple sclerosis, Alzheimer’s disease and other neurodegenerative conditions with myelin involvement, and even age-related cognitive decline,” adds Tara L. Moore, Ph.D., professor of anatomy and neurobiology.