Modified block uniform resampling (BURS) algorithm using truncated singular value decomposition: fast accurate...

Area of Science:

• Medical Imaging
• Signal Processing
• Magnetic Resonance Imaging (MRI)

Background:

• Nonuniform k-space sampling in MRI presents challenges for image reconstruction.
• The block uniform resampling (BURS) algorithm offers a simple, high-quality regridding solution but is sensitive to noise.
• Noise in k-space data can degrade image quality reconstructed with BURS.

Purpose of the Study:

• To present a modified BURS algorithm that mitigates noise sensitivity.
• To improve the robustness and performance of BURS for MRI reconstruction with noisy data.
• To reduce computational demands while enhancing image signal-to-noise ratio (SNR).

Main Methods:

• A modified BURS algorithm is proposed, incorporating singular value truncation after singular value decomposition (SVD).
• This technique selectively removes singular values to prevent noise amplification.
• The method aims to produce accurate gridded k-space data from noisy inputs.

Main Results:

• The modified BURS algorithm demonstrates reduced sensitivity to noise compared to the original BURS.
• Truncation of singular values effectively avoids amplification of noise from small singular values.
• The method leads to improved image SNR and reduced computational load.

Conclusions:

• Singular value truncation is an effective strategy to enhance the BURS algorithm's performance in the presence of noise.
• This modified approach offers a more robust and computationally efficient solution for MRI regridding.
• The technique holds promise for improving diagnostic image quality in clinical MRI settings.