Accelerating multi-echo T2 weighted MR imaging: analysis prior group-sparse optimization.

Area of Science:

• Medical Imaging
• Magnetic Resonance Imaging (MRI)

Background:

• Compressed Sensing (CS) reconstructs MRI from partial K-space data by exploiting spatial image correlation.
• Existing CS methods reconstruct individual T2 weighted images but do not fully utilize correlations between multiple echoes.

Purpose of the Study:

• To develop an advanced reconstruction technique for multi-echo T2 weighted MRI using partial K-space data.
• To improve image reconstruction by exploiting both intra-image spatial correlation and inter-image correlation across echoes.

Main Methods:

• Extended Compressed Sensing (CS) framework to incorporate inter-image correlation, formulating a group sparsity promoting optimization problem.
• Investigated group sparsity as both synthesis and analysis prior problems, focusing on the analysis prior for superior results.
• Evaluated reconstruction performance using Normalized Mean Squared Error (NMSE) under varying K-space sampling strategies.

Main Results:

• Group sparsity as an analysis prior, with appropriate sparsifying transforms, yielded superior reconstruction compared to standard sparse reconstruction.
• Significant improvements in reconstruction accuracy (lower NMSE) were observed when employing distinct K-space sampling patterns for each echo.
• No noticeable improvement from group sparsity was found when identical sampling patterns were used across all echoes.

Conclusions:

• The proposed group sparsity promoting framework significantly enhances the reconstruction of multi-echo T2 weighted MRI from undersampled data.
• The effectiveness of the group sparsity approach is critically dependent on the use of diverse sampling patterns for different echoes.
• This work offers a more robust and accurate method for accelerated MRI acquisition and reconstruction.