Improved least squares MR image reconstruction using estimates of k-space data consistency

Area of Science:

• Medical Imaging
• Data Reconstruction
• Magnetic Resonance Imaging (MRI)

Background:

• Magnetic resonance imaging (MRI) data can be corrupted by unfavorable magnetization evolution, leading to image quality degradation.
• Traditional reconstruction methods may not optimally handle noise and data inconsistencies simultaneously.

Purpose of the Study:

• To develop and evaluate a novel data reconstruction framework for corrupted MRI data.
• To improve image accuracy and precision by optimally balancing noise and data inconsistency.

Main Methods:

• A weighted least squares reconstruction approach using all available data and a data-derived consistency measure.
• Evaluation of algorithm variants in simulations, phantom experiments, and in vivo scans.
• Application to fast spin echo and self-gated respiratory gating techniques.

Main Results:

• The proposed data consistency weighting technique significantly enhanced image quality.
• A substantial reduction in image noise was observed compared to traditional reconstruction methods.
• The framework demonstrated compatibility with parallel imaging and compressed sensing applications.

Conclusions:

• The new weighted least squares reconstruction framework offers a significant improvement for corrupted MRI data.
• This method provides a robust approach for balancing noise and data inconsistencies, leading to superior image reconstruction.
• The technique shows promise for advancing various MRI applications requiring high-fidelity image estimation.