Optshrink LR + S: accelerated fMRI reconstruction using non-convex optimal singular value shrinkage

Area of Science:

• Medical Imaging
• Neuroimaging
• Signal Processing

Background:

• Accelerated Magnetic Resonance Imaging (MRI) is crucial for reducing scan times.
• Undersampled data in fMRI leads to artifacts and reduced image quality.
• Existing reconstruction methods struggle with balancing speed and accuracy.

Purpose of the Study:

• To introduce a novel accelerated fMRI reconstruction method named OptShrink LR+S.
• To enhance the reconstruction of undersampled fMRI data.
• To improve both qualitative and quantitative results compared to existing methods.

Main Methods:

• Reconstruction of undersampled fMRI data using a linear combination of low-rank and sparse components.
• Estimation of the low-rank component via a non-convex optimal singular value shrinkage algorithm.
• Estimation of the sparse component using convex l1 minimization.

Main Results:

• The OptShrink LR+S method demonstrated effective reconstruction of undersampled fMRI data.
• Qualitative and quantitative analyses showed superior performance compared to state-of-the-art algorithms.
• The method successfully balanced acceleration with high-fidelity image reconstruction.

Conclusions:

• The proposed OptShrink LR+S method offers a significant advancement in accelerated fMRI reconstruction.
• This technique holds promise for reducing fMRI scan times without compromising image quality.
• Further validation on diverse fMRI datasets is warranted.