A Simulation Framework for Quantitative Validation of Artefact Correction in Diffusion MRI.

Area of Science:

• Medical Imaging
• Neuroscience
• Computational Biology

Background:

• Diffusion-weighted magnetic resonance (DW-MR) imaging is crucial for brain microstructure and connectivity analysis.
• Image misalignment due to artifacts like eddy-current (EC) distortion and motion compromises data quality.
• Current validation methods for artifact correction are often qualitative or use surrogate metrics.

Purpose of the Study:

• To introduce a simulation framework for direct, quantitative comparison of DW-MR post-processing methods.
• To enable objective assessment and validation of existing and future artifact correction techniques.
• To identify systematic biases in commonly used correction methods.

Main Methods:

• Developed a physics-based simulation framework to generate realistic DW-MR datasets with inherent artifacts.
• Applied the framework to quantitatively assess a common eddy-current correction method (registration to b=0).
• Evaluated the impact of the correction method on image alignment and microstructural measures.

Main Results:

• The simulation framework accurately reproduces salient features and artifacts of DW-MR images.
• Testing revealed that registration of diffusion-weighted images (DWIs) to b=0 introduces systematic bias.
• This bias can compromise the accuracy of microstructural measures derived from corrected datasets.

Conclusions:

• The proposed simulation framework provides a robust platform for the objective validation of DW-MR image processing techniques.
• It facilitates direct, quantitative comparisons, moving beyond qualitative assessments.
• The findings highlight the need for rigorous validation of even standard correction methods to ensure reliable brain imaging analysis.