Geometric distortion correction of high-resolution 3 T diffusion tensor brain images

Area of Science:

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

Background:

• Echo planar imaging (EPI) is prone to geometric distortions caused by magnetic field inhomogeneities and eddy currents.
• These distortions compromise the anatomical accuracy of diffusion-weighted images, impacting applications like diffusion tensor imaging (DTI).
• Accurate correction of EPI distortions is crucial for reliable quantitative analysis and clinical interpretation.

Purpose of the Study:

• To present a novel nonlinear warping approach using optic flow for correcting geometric distortions in EPI acquired at 3 Tesla.
• To improve the anatomical fidelity of baseline and diffusion-weighted EPI sequences.
• To enable the generation of accurate diffusion tensor images at 3T.

Main Methods:

• A nonlinear warping algorithm based on optic flow was developed to correct EPI distortions.
• Echo planar images (EPI) were warped to align with anatomically correct T2-weighted fast spin echo (T2-FSE) images.
• Intensity matching (global histogram matching for T2-FSE and local matching for EPI) was employed prior to warping.
• Distortion correction was evaluated using visual contour overlays, global mutual information, and a local distance measure.

Main Results:

• The optic flow-based warping algorithm successfully reduced geometric distortions in EPI.
• Visual assessment and global mutual information indicated significant improvement in image alignment.
• The local distance measure demonstrated that distortions were corrected to sub-voxel levels.
• The method proved effective for both baseline and diffusion-weighted EPI at 3 Tesla.

Conclusions:

• The presented nonlinear warping algorithm effectively corrects geometric distortions in 3T EPI.
• This technique facilitates the generation of anatomically accurate diffusion tensor images.
• The approach holds promise for enhancing the reliability of diffusion MRI studies.