A model-based reconstruction for undersampled radial spin-echo DTI with variational penalties on the diffusion...

Area of Science:

• Medical Imaging
• Biophysics
• Diffusion Tensor Imaging

Background:

• Radial spin-echo diffusion imaging offers motion-robustness for low T2 tissues like articular cartilage.
• Challenges in in vivo measurements include slow acquisition and inefficient k-space coverage, necessitating accelerated protocols via undersampling.

Purpose of the Study:

• Introduce a novel model-based reconstruction approach for undersampled radial spin-echo diffusion tensor imaging (DTI).
• Improve image quality and reduce acquisition time for DTI without compromising parameter quantification or signal-to-noise ratio (SNR).

Main Methods:

• A model-based reconstruction exploiting image redundancies to reduce unknowns.
• Compressed sensing applied directly in the quantitative domain using a total variation (TV) constraint on diffusion tensor elements.
• Experiments conducted on an anisotropic phantom and human knee and brain datasets.

Main Results:

• The proposed method demonstrated reduced noise and streaking artifacts in quantitative parameter maps compared to existing methods.
• Observed a reduction in angular dispersion of the primary eigenvector.
• No systematic errors were introduced into the parameter maps.

Conclusions:

• The new reconstruction approach enables significant acquisition time reduction in radial spin-echo DTI.
• Maintains or improves parameter quantification and SNR.
• Offers a promising solution for accelerated in vivo DTI of challenging tissues.