Normalized gradient fields for nonlinear motion correction of DCE-MRI time series

Area of Science:

• Medical Imaging
• Image Processing
• Biomedical Engineering

Background:

• Dynamic contrast-enhanced MRI (DCE-MRI) captures organ motion and contrast agent dynamics.
• Motion artifacts in DCE-MRI hinder accurate physiological parameter assessment.
• Deformable image registration is crucial for correcting motion in DCE-MRI.

Purpose of the Study:

• To present a novel partial differential equation-based method for deformable multimodal image registration.
• To address motion challenges in DCE-MRI time series, particularly for moving organs.
• To compare the performance of normalized gradients and mutual information as cost functionals for motion correction.

Main Methods:

• A partial differential equation-based method using normalized gradients and Fourier transform.
• Solving Euler-Lagrange equations within a multilevel hierarchy.
• Validation on ten DCE-MRI datasets from moving kidney studies.

Main Results:

• Both normalized gradients and mutual information proved effective for DCE-MRI motion correction.
• Normalized gradients demonstrated superior performance over mutual information based on multiple metrics.
• The method was validated on real-world DCE-MRI data from moving kidneys.

Conclusions:

• Normalized gradients are a viable and accurate alternative to mutual information for DCE-MRI registration.
• The proposed method shows promise for clinical applications involving DCE-MRI of moving organs.
• Accurate motion correction enhances the reliability of physiological parameter assessment from DCE-MRI data.