Referenceless reduction of spin-echo echo-planar imaging distortion with generative displacement mapping.

Area of Science:

• Medical Imaging
• Neuroimaging
• Computational Neuroscience

Background:

• Echo-planar imaging (EPI) is crucial for diffusion MRI but susceptible to geometric distortions.
• Retrospective studies often lack necessary reference scans (field maps or reversed-gradient scans) for distortion correction.
• Accurate EPI data is vital for reliable diffusion index calculation and anatomical registration.

Purpose of the Study:

• To develop a fully automatic, referenceless method for correcting EPI distortions.
• To enable distortion correction in retrospective studies lacking reference field maps or reversed-gradient scans.
• To specifically address EPI data acquired with anterior-posterior or posterior-anterior phase-encoding.

Main Methods:

• Utilized a generative adversarial network (GAN) to generate a displacement map.
• Inputted a 3D raw b0 volume from diffusion-tensor data into the GAN.
• Applied the generated displacement map to correct EPI images across the entire diffusion dataset.

Main Results:

• Demonstrated effective reduction of geometric distortions in EPI data sets.
• Showed improved accuracy of diffusion indices.
• Significantly enhanced coregistration between EPI and high-resolution T1-weighted images (p < 0.01).

Conclusions:

• Developed a practical, referenceless EPI distortion correction method.
• Publicly released the method as a standalone application.
• Provides a valuable tool for retrospective studies to improve EPI data quality and diffusion measure accuracy.