New Method Maps Myelin Patterns Across the Brain Automatically

Myelin, the fatty sheath that insulates nerve fibers and keeps electrical signals moving quickly through the brain, doesn’t look the same everywhere. Its patterns shift dramatically from one brain region to the next, and until now there hasn’t been a good way to measure that variation directly. Researchers at the University of Eastern Finland have built a computer method that quantifies these myelin patterns automatically, without requiring anyone to manually label the tissue images first. The work appears in Brain Structure and Function. 

Measuring myelin patterns has traditionally meant staining tissue sections and then assessing them by eye, region by region, or relying on simple quantitative measures such as staining intensity—approaches that miss much of what’s actually going on in the tissue. “Myelin covers most axons in the brain and altogether forms such a variety of patterns. Some brain areas are densely packed with aligned myelinated axons such as the corpus callosum and optic nerve. Others, like the cortex, vary in density and orientation, from axons running in a clear direction to net-like patterns, with every intermediate case in between. We needed a way to quantify those patterns properly, not just describe them by eye, or reduce them to basic measures like staining intensity,” says first author Melina Estela.

The new method works by splitting each tissue image into small windows, then using a convolutional autoencoder to compress the information within each window down to a small set of numbers. Windows that produce similar numbers are grouped together, generating unsupervised maps that sort the tissue into regions with distinct properties on their own—separating white matter from grey matter at the broadest level, and picking out finer structures like the hippocampal subfields and cortical layers at a more detailed one.

To test what the maps could reveal, the researchers trained the method on tissue images from both healthy and injured animals and compared how the resulting quantification differed between the two groups. One map, corresponding to major white matter tracts, came out reduced in the injured animals, a change that lines up directly with axonal damage. Other maps also showed differences in proportion between healthy and injured tissue, hinting at additional, subtler changes that will require further study to fully interpret. 

Because the method doesn’t rely on any manual labeling, it could make large-scale studies of brain tissue both faster and more consistent across labs. The researchers note the approach isn’t limited to myelin, either—it could be adapted to other stains, tissue types, and diseases.