Direct diffusion tensor estimation using a model-based method with spatial and parametric constraints.

Area of Science:

• Medical Imaging
• Biophysics
• Computational Science

Background:

• Diffusion Tensor Imaging (DTI) is crucial for characterizing biological tissues.
• Accelerating DTI acquisition is essential for reducing scan times and motion artifacts.
• Existing acceleration methods often compromise image quality and accuracy.

Purpose of the Study:

• To develop and validate a novel model-based method with spatial and parametric constraints (MB-SPC) for accelerated DTI.
• To directly estimate diffusion tensors from highly undersampled k-space data.
• To improve the accuracy and reduce artifacts in accelerated DTI.

Main Methods:

• The MB-SPC method utilizes joint sparsity (L1-L2 norm) and tensor smoothness (total variation seminorm) priors.
• Undersampled k-space data were acquired from simulated phantoms and ex-vivo rat hearts (acceleration factors 2-4).
• Diffusion tensors were reconstructed using a nonlinear conjugate gradient descent algorithm, with performance assessed by normalized root mean square error (nRMSE).

Main Results:

• MB-SPC achieved acceptable DTI measures at acceleration factors up to 4.
• The method demonstrated more accurate diffusion tensor estimation compared to existing techniques at higher acceleration factors.
• Quantitative assessment showed reduced nRMSE for DTI indices.

Conclusions:

• The MB-SPC method significantly reduces artifacts in accelerated DTI, particularly at higher acceleration factors or lower signal-to-noise ratios (SNRs).
• This technique is adaptable for MR relaxometry parameter mapping.
• It holds promise for characterizing biological tissues like nerves, muscle, and heart tissue.