Correcting Susceptibility-Induced Distortion in Diffusion-Weighted MRI using Constrained Nonrigid Registration.

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Neuroimaging
• Image Processing

Background:

• Echo Planar Imaging (EPI) is standard for diffusion MRI but suffers from susceptibility-induced B0 field inhomogeneities.
• These inhomogeneities cause geometric distortion in brain diffusion images, hindering anatomical correspondence.
• Accurate anatomical alignment is crucial for interpreting diffusion MRI data.

Purpose of the Study:

• To develop and test a novel approach for estimating and correcting geometric distortion in diffusion-weighted MRI.
• To improve the spatial accuracy of diffusion MRI scans of the human head.
• To enable better correspondence between diffusion and anatomical MRI datasets.

Main Methods:

• A non-linear registration framework utilizing mutual information was employed.
• The commonly acquired anatomical MRI scan served as the registration template.
• Registration was constrained by spatial regularization and physics-based distortion characteristics, avoiding extra data acquisition.

Main Results:

• The proposed method successfully estimated and corrected geometric distortion in simulated diffusion MRI data.
• Experimental data demonstrated the effectiveness of the distortion correction technique.
• The corrected diffusion images showed improved spatial alignment with anatomical references.

Conclusions:

• The developed non-linear registration approach effectively corrects EPI-induced geometric distortion in diffusion MRI.
• This method enhances the anatomical accuracy of diffusion MRI without requiring additional imaging sequences or data.
• The findings facilitate more reliable integration of diffusion and anatomical MRI for neuroimaging research.