Unified NDMI for small intestinal cine MRI: repeatability and sensitivity to multicenter variability

Area of Science:

• Medical Imaging
• Gastrointestinal Physiology
• Quantitative MRI

Background:

• Assessing small intestinal motility with cine-MRI is crucial for diagnosing gastrointestinal disorders.
• Existing methods for analyzing cine-MRI data lack standardization, hindering comparability across studies and centers.
• A unified, feature-agnostic approach is needed to enable reliable quantitative assessment of intestinal motility.

Purpose of the Study:

• To introduce a novel, comparability-first cine-MRI paradigm for small intestinal motility analysis.
• To establish a unified, feature-agnostic normalized differential motion index (NDMI) for quantitative motility assessment.
• To investigate scan-rescan consistency and sensitivity to simulated multicenter variability across Optical flow, Morphology, and Implicit Neural Representation (INR) methods.

Main Methods:

• Twelve healthy volunteers underwent same-scanner scan-rescan cine-MRI.
• Motility features were extracted using generalized Optical flow, JointSwinUNETR segmentation (Morphology), and INR registration.
• A single NDMI parameterization was applied across methods, compared with expert ratings, and assessed for repeatability and robustness to perturbations.

Main Results:

• All three methods demonstrated favorable agreement with expert ratings, with INR achieving the highest accuracy and Morphology yielding the strongest weighted kappa.
• Morphology exhibited the tightest limits of agreement in Bland-Altman analyses, indicating superior repeatability, while INR showed good robustness to affine changes.
• Dominant variance sources differed across methods under simulated multicenter variability, with Morphology being the fastest (0.03s/frame).

Conclusions:

• The normalized differential motion index (NDMI) provides a unified, feature-agnostic basis for comparing intestinal motility MRI methods.
• Morphology demonstrated the highest consistency, INR showed the best expert agreement and robustness, and Optical flow was more sensitive to signal-to-noise ratio.
• These findings support scenario-aware method selection and advocate for prospective multicenter validation with protocol harmonization for small intestinal motility MRI.