Compressed sensing parallel magnetic resonance imaging.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Signal Processing

Background:

• Magnetic Resonance Imaging (MRI) is crucial for diagnostics but suffers from long scan times.
• Parallel Magnetic Resonance Imaging (pMRI) and Compressed Sensing (CS) are techniques to accelerate MRI acquisition.
• Integrating pMRI and CS can leverage coil sensitivity and image sparsity for improved reconstruction.

Purpose of the Study:

• To develop and evaluate an integrated pMRI and CS method for accelerated MRI.
• To utilize CS as a regularization technique for the inverse problem in pMRI.
• To assess the performance of the integrated method against existing regularization techniques.

Main Methods:

• The study integrates pMRI with CS, using L1 norm and Total Variation (TV) as regularization terms.
• The method addresses the inverse problem arising from pMRI reconstruction.
• In-vivo 8-channel brain MRI data were used for testing at various reduction factors (2-8).

Main Results:

• The proposed integrated pMRI and CS method demonstrated superior performance compared to truncated Singular Value Decomposition (SVD) and Tikhonov regularization.
• The method effectively reduced residual artifacts and improved Signal-to-Noise Ratio (SNR).
• Performance benefits were particularly notable at reduction factors exceeding 4.

Conclusions:

• The integration of pMRI and CS offers a powerful approach for accelerating MRI acquisition while maintaining high image quality.
• The proposed CS-based regularization method significantly outperforms traditional techniques for parallel MRI reconstruction.
• This technique holds promise for faster and more efficient in-vivo MRI, especially for brain imaging.