DISTORTION-OPTIMAL SELF-CALIBRATING PARALLEL MRI BY BLIND INTERPOLATION IN SUBSAMPLED FILTER BANKS

Area of Science:

• Medical Imaging
• Biophysics
• Signal Processing

Background:

• Parallel MRI accelerates imaging by using multiple receiver coils.
• Current methods often require extensive calibration data, limiting acceleration.
• Artifacts can arise from undersampled data in k-space.

Purpose of the Study:

• To develop a self-calibrating k-space-based image reconstruction method for parallel MRI.
• To enable high effective acceleration with minimal calibration data.
• To improve image quality and reduce scan times.

Main Methods:

• Utilized a filter bank interpolation framework for k-space data.
• Formulated a new inverse problem for blind identification of interpolant filters.
• Applied the method to scenarios requiring high acceleration.

Main Results:

• Successfully reconstructed artifact-free images using minimal calibration data.
• Achieved high effective acceleration factors.
• Demonstrated improved image quality and scan time reduction over GRAPPA.

Conclusions:

• The proposed self-calibrating method enhances parallel MRI reconstruction.
• It offers superior performance compared to state-of-the-art techniques like GRAPPA.
• This approach facilitates significant reductions in MRI scan times.