Automated pipeline normalizes myelin staining for cross-modal MRI validation
New algorithm removes staining noise, enabling precise myelin quantification at scale.
A team of 24 researchers from the University of Pennsylvania, led by Zahra Khodakarami, has introduced an automated pipeline that normalizes Optical Density (OD) in Luxol Fast Blue histopathology for quantitative myelin analysis. The core problem: staining variability across lab protocols introduces systematic noise that obscures true myelin concentration. The pipeline automatically detects non-pathologic reference regions in whole-slide images to compute normalized OD heatmaps, dramatically reducing artifacts.
Validation was two-pronged: first, the reference selection matched expert-identified regions with high concordance. Second, cross-modal comparison with co-registered 7T ex vivo MRI showed normalized OD correlated significantly stronger with MRI signal intensity than raw OD. Notably, the correlation held within white matter hyperintensities (WMH), proving the pipeline captures continuous myelin pathology, not just binary presence of loss. Accepted at MICCAI 2026, this work establishes a robust foundation for future in vivo myelin mapping and biomarker discovery.
- Automated reference region selection eliminates staining variability across histology runs.
- Normalized OD shows stronger correlation with 7T MRI signal than raw OD measurements.
- Pipeline validated on continuous myelin pathology within white matter hyperintensities, not just healthy tissue.
Why It Matters
Enables reliable cross-modal myelin quantification, advancing biomarker discovery for neurodegenerative diseases like multiple sclerosis and Alzheimer's.