HASAN: Highly accurate sensitivity for auto-contrast-corrected pMRI reconstruction

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Image Reconstruction
• Signal Processing

Background:

• Accurate coil sensitivity map estimation is crucial for high-quality MRI reconstruction.
• Existing methods often require prior information or simultaneous object image estimation.
• Self-calibrating SENSE reconstruction demands precise coil sensitivity maps.

Purpose of the Study:

• To propose a novel, highly accurate coil sensitivity-map estimation method.
• To improve image reconstruction in MRI using self-calibrating SENSE.
• To achieve auto-corrected contrast in reconstructed images without post-processing.

Main Methods:

• Developed a constrained image-domain multi-channel LMS (c-iMCLMS) algorithm.
• Utilized low-resolution images from variable density MR data to form an image-domain cross-relation equation.
• Solved the equation iteratively using a novel sum-of-squares (SOS) constraint and SOS normalization for faster convergence.
• Employed a regularized conjugate gradient (CG) based SENSE algorithm for image reconstruction.

Main Results:

• The c-iMCLMS algorithm demonstrated highly accurate coil sensitivity-map estimation.
• Achieved significant signal-to-artifact-noise-ratio (SANR) improvement compared to existing techniques.
• Reconstructed true object images with auto-corrected contrast, eliminating the need for post-contrast correction.
• Validated the method on simulation, synthetic, and in-vivo datasets.

Conclusions:

• The proposed c-iMCLMS method offers a robust and accurate approach for coil sensitivity-map estimation in MRI.
• This technique enhances image reconstruction quality and contrast correction.
• The method provides significant improvements in SANR, approaching theoretical limits.