Post

There are many MRI approaches for studying myelin. But which ones are reliable, how do they relate to each other, and how do they behave in gray vs white matter? In our new preprint, we introduce Myelin Imaging Reliability and ReprOducibility Resource (MIRROR) 🧠🪞 📄🔗 www.biorxiv.org/content/10.6...

MIRROR includes 2 MRI sessions in 22 adults, with: - diffusion MRI (dMRI) - ihMT - MP2RAGE - QSM - T1w/T2w - Relaxometry We derive 27 myelin-sensitive metrics for primary analyses (93 metrics in the supplement!).

First question: how well do these metrics differentiate gray and white matter? The answer: a lot! MP2RAGE R1 showed the strongest differentiation, followed by several dMRI, T1w/T2w, and ihMT measures. QSM measures showed much weaker and more heterogeneous differences.

How similar are different myelin-sensitive measures to each other? Intermetric correlations were generally stronger in white matter and strongest among measures from the same acquisition/modeling family. QSM measures were particularly distinct from most of the other metrics.

Next: test–retest reliability. dMRI-derived metrics were consistently among the most reliable measures, with many median ICCs approaching or exceeding 0.9. MP2RAGE R1 and T1w/T2w also performed well. QSM, R2*, and ihMT (especially in gray matter) were more variable.

We also asked whether each metric preserves a participant-specific spatial “fingerprint” across sessions. Most metrics had extremely high discriminability, even when local ICCs were more variable.

Takeaway #1: There is no single interchangeable myelin-sensitive MRI measure. Metrics differ in tissue contrast, reliability, and relationships with one another, and these relationships can depend on whether you are studying gray or white matter.

Takeaway #2: Tissue context matters. Relationships among myelin-sensitive metrics were generally stronger in white matter and were not always preserved in cortical gray matter. Findings from one tissue compartment may not generalize to the other.

Takeaway #3: Reliable does not mean equivalent. Many metrics showed strong test–retest reliability, but they differed substantially in tissue contrast and in how they related to one another. Reliability alone does not tell us whether two measures capture the same biology.

We hope MIRROR can be used to support new method development, multimodal integration, and validation of MRI-derived myelin measures against histology and other independent references.

MIRROR is completely open: 🧠 Raw MRI data 🗺️ Derived metric maps 📊 Analysis derivatives 💻 Processing + analysis code Data: openneuro.org/datasets/ds0... Code: github.com/PennLINC/mir...