ESPIRiT--an eigenvalue approach to autocalibrating parallel MRI: where SENSE meets GRAPPA

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Image Reconstruction
• Parallel Imaging

Background:

• Parallel imaging reconstructs MRI data using undersampled multicoil acquisitions.
• Key methods include SENSE (Sensitivity Encoding) and GRAPPA (Generalized Autocalibrating Partially Parallel Acquisitions).
• These methods differ in their approach to utilizing coil sensitivities and k-space correlations.

Purpose of the Study:

• To elucidate the theoretical relationship between SENSE and GRAPPA.
• To develop and validate an improved parallel imaging algorithm.
• To bridge the gap between existing SENSE and GRAPPA methodologies.

Main Methods:

• Theoretical analysis of k-space correlations and coil sensitivities.
• Investigated the null space of calibration matrices and solution subspaces.
• Developed an extended SENSE reconstruction using multiple sensitivity maps.

Main Results:

• Confirmed k-space correlations are encoded in the null space of calibration matrices.
• Demonstrated that SENSE and GRAPPA solutions are restricted to subspaces spanned by coil sensitivities.
• The extended SENSE reconstruction integrated SENSE advantages with GRAPPA-like robustness.

Conclusions:

• The theoretical gap between SENSE and GRAPPA has been bridged.
• A novel autocalibration technique effectively combines the benefits of both SENSE and GRAPPA.
• This new approach offers improved image reconstruction for undersampled multicoil MRI data.