Evaluating efficient SENSE algorithms to deblur spiral MRI with fat/water separation

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Medical Imaging Reconstruction
• Computational Efficiency in Imaging

Background:

• SENSE and spiral imaging with fat/water separation offer high temporal efficiency in MRI.
• However, multi-channel data processing increases computational complexity due to blurring/deblurring operations.

Purpose of the Study:

• To present and evaluate two alternative models for simplifying computational complexity in spiral MRI reconstruction.
• To assess the performance of these models in terms of computation time and reconstruction error.

Main Methods:

• Two approximated spiral MRI reconstruction models (Model 2 and Model 3) were developed by altering the order of the coil-sensitivity encoding process.
• These models were tested on T1- and T2-weighted brain image data from four subjects, using simulated undersampling.

Main Results:

• Model 2 reduced computation time by 31%-47%, and Model 3 by 39%-56%, compared to the full model (Model 1).
• Water image quality was consistent across models; Model 3 fat images matched Model 1, while Model 2 showed up to 4.8% higher normalized error.

Conclusions:

• Model 2 offers the fastest computation but introduces higher fat channel error, especially at high fields or with long acquisition windows.
• Model 3 provides a faster, accurate alternative to the full model, maintaining high reconstruction quality.