Autocalibrating segmented diffusion-weighted acquisitions.

Area of Science:

• Magnetic Resonance Imaging
• Image Reconstruction

Background:

• Segmented echo-planar imaging (EPI) is crucial for high-resolution diffusion-weighted imaging (DWI).
• Phase differences between segments in segmented EPI can cause significant image artifacts, limiting diagnostic quality.

Purpose of the Study:

• To develop and validate an algorithm for reconstructing interleaved segmented DWI acquisitions.
• The algorithm aims to eliminate the need for additional calibration or navigator measurements to correct for phase inconsistencies.

Main Methods:

• A parallel imaging algorithm is proposed that jointly reconstructs all segments of a DWI frame, preserving phase information.
• Iterative refinement of phase estimates is incorporated into the joint reconstruction process.
• Initial phase estimates are derived from traditional parallel imaging reconstruction using unweighted scans.

Main Results:

• Phantom data reconstruction and g-factor simulations demonstrated significant improvements, with up to a 93% reduction in root mean square error compared to existing methods.
• In vivo experiments confirmed robust reconstruction performance even with limited receiver channels or low signal-to-noise ratios.

Conclusions:

• A robust algorithm for segmented DWI data reconstruction has been developed.
• The method obviates the need for navigator or calibration scans, making it applicable to retrospective analysis of existing DWI datasets.