SENSE with improved tolerance to inaccuracies in coil sensitivity maps

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Image Reconstruction
• Parallel Imaging

Background:

• The standard Cartesian sensitivity encoding (SENSE) framework optimizes for signal-to-noise ratio (SNR) in parallel MRI.
• Artifacts can arise in SENSE reconstructions due to inaccuracies in receiver coil sensitivity knowledge.

Purpose of the Study:

• To propose an extension of the Cartesian SENSE framework that accounts for uncertainties in receiver coil sensitivities.
• To balance artifact reduction with signal-to-noise ratio preservation in accelerated MRI.

Main Methods:

• Introduced an extension to the SENSE framework incorporating receiver coil sensitivity uncertainty.
• Weighted the inversion problem to balance artifact levels and SNR using a least-squares optimization.
• Assessed performance using abdominal imaging data.

Main Results:

• Substantially reduced artifact levels in abdominal MRI reconstructions.
• Observed a signal-to-noise ratio (SNR) penalty, the magnitude of which is dependent on sensitivity inaccuracy, acceleration factor, and coil configuration.
• The method avoids complex iterative schemes, utilizing least-squares optimization.

Conclusions:

• The proposed extension of the SENSE framework effectively reduces artifacts in parallel MRI.
• The trade-off between artifact reduction and SNR is controllable by adjusting parameters related to coil sensitivity accuracy and imaging configuration.