Methods for quantitative susceptibility and R2* mapping in whole post-mortem brains at 7T applied to amyotrophic lateral sclerosis

Chaoyue Wang, Sean Foxley, Olaf Ansorge, Sarah Bangerter-Christensen, Mark Chiew, Anna Leonte, Ricarda A. Menke, Jeroen Mollink, Menuka Pallebage-Gamarallage, Martin R. Turner, Karla L. Miller, Benjamin C. TendlerNeuroImage 2020 · DOI

TL;DR

This paper introduces a robust pipeline for quantitative susceptibility mapping (QSM) and R2* mapping in fixed, whole post-mortem human brains. They demonstrate its utility by comparing these MRI measures to iron and myelin histology in brains from individuals with amyotrophic lateral sclerosis (ALS) and controls, revealing potential MRI biomarkers for neurodegenerative diseases.

What Problem Does This Paper Address?

Magnetic Resonance Imaging (MRI) offers valuable insights into brain tissue, but its biological specificity can be limited. Susceptibility Weighted Imaging (SWI) techniques like QSM and R2 mapping are sensitive to iron and myelin, both implicated in neurodegenerative diseases like ALS. However, applying these techniques to whole, fixed post-mortem brains presents unique challenges: air bubbles can cause signal dropout, brain shape can affect magnetic field homogeneity, and the presence of fixatives can alter tissue properties. Furthermore, directly correlating MRI findings with detailed histology in the same tissue requires specialized methods. This paper aims to bridge this gap by developing and validating a comprehensive pipeline for quantitative susceptibility and R2 mapping in whole, fixed post-mortem brains, specifically applied to ALS research.

Methods

The authors developed a custom image processing pipeline to address the challenges of post-mortem brain imaging. Key steps include:

  • Custom Reconstruction: A pipeline to reconstruct raw k-space data, removing coil-specific phase offsets and aligning repeat scans to eliminate artifacts like "fringe lines" and image misalignments caused by magnetic field drift.
  • Phase Unwrapping and Field Map Estimation: A robust method for estimating the magnetic field perturbation (ΔB₂) and phase offset (Φ₀) from multi-echo gradient echo (GRE) data, using a non-linear fit and careful initialization to avoid phase wraps.
  • Background Field Removal and Dipole Inversion: Evaluation and selection of the most suitable algorithms for post-mortem data. They found that a combination of v-SHARP for background field removal and STAR-QSM for dipole inversion yielded the most robust susceptibility maps, minimizing streaking artifacts and large-scale inhomogeneities.
  • R2* Mapping: A voxel-wise fitting of magnitude images from all echoes to estimate R2* values.
  • Histological Analysis: Semi-quantitative analysis of ferritin (iron storage protein) and myelin proteolipid protein (PLP) using immunohistochemistry on tissue samples from specific brain regions (primary motor cortex (M1), anterior cingulate cortex (ACC), and visual cortex (V2)).
  • Correlation and Comparison: Statistical analysis to correlate MRI measures (susceptibility and R2*) with histological findings and to compare ALS brains with control brains.

Datasets

The study utilized a dataset of twelve whole, post-mortem formalin-fixed human brains. This cohort consisted of nine donors with a clinical diagnosis of ALS and three control donors with no known neurological disorder. The brains were scanned at 7 Tesla (7T) using a multi-echo GRE sequence.

Key Insights and Ideas

  • Whole-Brain QSM/R2* is Feasible: The paper demonstrates that it's possible to obtain high-quality quantitative susceptibility and R2* maps from entire fixed post-mortem brains, overcoming significant technical hurdles.
  • MRI-Histology Correlation is Crucial: Directly comparing MRI measures to histological stains in the same tissue is essential for understanding the biological basis of MRI signals, especially in disease contexts.
  • ALS Brains Show Altered Susceptibility and R2*: The study found elevated magnetic susceptibility and R2* values in the primary motor cortex (M1) of ALS brains compared to controls.
  • Iron and Myelin Link to MRI Signals: Both susceptibility and R2* values showed positive correlations with ferritin and myelin estimates from histology, suggesting these are key contributors to the MRI contrast.
  • Variability in ALS Pathology: Not all ALS brains showed the same degree of hyperintensity in M1, highlighting potential heterogeneity in disease presentation or progression.

Key Results

  • Increased M1 Values in ALS: ALS brains showed significantly higher mean susceptibility and R2 values in the primary motor cortex (M1) compared to control brains (p=0.0098 for susceptibility, p=0.0065 for R2).
  • Positive Correlations: Magnetic susceptibility and R2 values demonstrated positive correlations with both ferritin and myelin (PLP) estimates from histology across different brain regions (e.g., R=0.73 for R2 vs. susceptibility).
  • Visual Hyperintensities: Four out of nine ALS brains exhibited visually apparent hyperintense susceptibility and R2* values in the cortical grey matter of M1.
  • Fixative Effects: Different fixatives (10% NBF vs. 10% formalin) had a significant impact on R2* values (lower in NBF) but not on susceptibility values in M1.

Limitations and Future Work

The authors acknowledge several limitations:

  • Registration Challenges: Direct pixel-wise correlation between MRI and histology was not possible due to the lack of automated registration between histology slides and 3D MRI data. Future work aims to develop such a pipeline.
  • Ferritin Staining Variability: Ferritin staining quality can be operator-dependent, leading to batch effects. Standardizing SAFs within batches helped, but future studies might benefit from single-batch processing or more robust quantification methods.
  • QSM Reference Region: Defining a truly "pathology-free" reference region for QSM in post-mortem brains is difficult. The study used the whole brain mean as a reference, which is a common approach but has limitations.
  • Small Control Group: The study included only three control brains, limiting statistical power for group comparisons.
  • Fixative Diffusion Effects: Variations in R2* values with distance from the tissue surface were observed, potentially due to fixative diffusion. A regression-based correction is being developed.

Future work will focus on improving registration for pixel-wise comparisons, refining histological quantification, and expanding the control group to further validate the pipeline and its findings in neurodegenerative diseases.

Summary updated

BibTeX
@article{wang2020methods,
  author = {Chaoyue Wang and Sean Foxley and Olaf Ansorge and Sarah Bangerter-Christensen and Mark Chiew and Anna Leonte and Ricarda A. Menke and Jeroen Mollink and Menuka Pallebage-Gamarallage and Martin R. Turner and Karla L. Miller and Benjamin C. Tendler},
  title = {{Methods for quantitative susceptibility and R2* mapping in whole post-mortem brains at 7T applied to amyotrophic lateral sclerosis}},
  journal = {NeuroImage},
  year = {2020},
  doi = {10.1016/j.neuroimage.2020.117216},
}
Figure