Quantitative Semisolid Magnetization Transfer and Relayed Nuclear Overhauser Effect Imaging in a Multiple Sclerosis Mouse Model Using Deep Magnetic Resonance Fingerprinting

Technical Summary: Quantitative Semisolid Magnetization Transfer and Relayed Nuclear Overhauser Effect Imaging in a Multiple Sclerosis Mouse Model Using Deep Magnetic Resonance Fingerprinting

Problem Statement
Magnetic resonance imaging (MRI) remains the primary modality for diagnosing, characterizing, and monitoring multiple sclerosis (MS). However, a significant limitation exists: the contrasts produced by MS lesions frequently overlap with those of other pathological conditions, necessitating the identification of more specific disease biomarkers. While saturation transfer (ST) MRI offers a pathway to access molecular information regarding myelin, proteins, and lipids, the accurate quantification of the underlying proton exchange parameters has historically been challenging.

Methodology
To address these challenges, the authors developed a strategy extending and modifying AI-boosted ST magnetic resonance fingerprinting (MRF) imaging at a 7T field strength. The study utilized a longitudinal cuprizone-induced MS mouse model (n=12) to evaluate the technique's efficacy. The primary objective was to quantify the dynamics of two specific proton pools:

1. Semisolid Magnetization Transfer (MT): Associated with the solid-like macromolecular environment.
2. Relayed Nuclear Overhauser Effect (rNOE): Specifically targeting aliphatic protons at chemical shifts of -3.5 ppm and -1.6 ppm relative to water.

The methodology involved reconstructing proton volume fractions from the MRF data. Validation was performed in two stages:

• In Vitro: Using lipid phantoms to correlate reconstructed proton volume fractions with known lipid concentrations.
• In Vivo: Monitoring the corpus callosum over time to detect changes in semisolid MT and rNOE parameters during cuprizone feeding. These findings were subsequently compared against conventional water relaxometry and histological analysis.

Key Contributions
The paper introduces a modified, AI-enhanced ST-MRF framework capable of rapid, multi-pool quantification. A central contribution is the successful application of this technique to simultaneously resolve semisolid MT and rNOE dynamics in a living animal model. The study demonstrates that this approach can extract specific molecular information related to myelin and lipid content that is distinct from standard water-based relaxometry.

Results

• Phantom Validation: In lipid phantoms, the reconstructed proton volume fractions showed a strong correlation with known lipid concentrations across all three proton pools (semisolid MT, rNOE at -3.5 ppm, and rNOE at -1.6 ppm), with correlation coefficients greater than 0.96 (p<0.001).
• In Vivo Findings: In the corpus callosum of the mouse model, both semisolid MT and rNOE proton volume fractions demonstrated a statistically significant decrease (p<0.01) as early as week 4 of cuprizone feeding.
• Comparative Sensitivity: These molecular changes were detected earlier than alterations observed via conventional water relaxometry.
• Histological Correlation: The biomarkers derived from ST-MRF were found to be in agreement with histological findings, confirming the biological relevance of the quantitative metrics.

Significance
The authors conclude that their results demonstrate the feasibility of using rapid, multi-pool ST-MRF quantification for the characterization of MS. By providing a method to quantify specific proton exchange parameters associated with myelin and lipids earlier than conventional techniques, this approach offers a potential avenue for improving the specificity of MS diagnosis and monitoring, addressing the current limitation of overlapping lesion contrasts in standard MRI.