Double total variation (DTV) regularization and Improved adaptive moment estimation (IADAM) optimization method for...

Area of Science:

• Medical Imaging
• Image Reconstruction
• Signal Processing

Background:

• Compressed sensing is crucial for fast Magnetic Resonance (MR) image reconstruction.
• Traditional total variation methods often result in staircase artifacts due to limited subband analysis.
• Existing algorithms struggle with artifacts in high-acceleration MR imaging.

Purpose of the Study:

• To introduce a novel algorithm, Double Total Variation (DTV), to mitigate staircase effects in MR image reconstruction.
• To enhance image quality by increasing variational information from diagonal subbands.
• To improve the efficiency and accuracy of fast MR image reconstruction.

Main Methods:

• Developed the Double Total Variation (DTV) regularization technique incorporating diagonal subband information.
• Implemented an Improved Adaptive Moment Estimation (IADAM) algorithm to optimize the conjugate gradient method.
• Validated the DTV algorithm using MR images of the head, knee, and ankle under various acceleration factors.

Main Results:

• The proposed DTV algorithm combined with IADAM conjugate gradient demonstrated superior reconstruction performance compared to traditional methods.
• Simulations across multiple body parts and acceleration factors confirmed the general superiority of the DTV approach.
• Significant reduction in staircase artifacts was observed in reconstructed MR images.

Conclusions:

• The Double Total Variation (DTV) method effectively reduces staircase artifacts in MR image reconstruction.
• The proposed algorithm offers a promising advancement for fast MR imaging techniques.
• This work provides a foundation for further algorithmic improvements in the field of image reconstruction.