Model-based reconstruction of undersampled diffusion tensor k-space data.

Area of Science:

• Medical Imaging
• Neuroimaging
• Biophysics

Background:

• Diffusion Tensor Imaging (DTI) is valuable for high-resolution 3D imaging of ex vivo specimens.
• Long acquisition times currently limit the practical utility of DTI, especially for complex experiments.

Purpose of the Study:

• To develop and evaluate a compressed sensing (CS) framework to accelerate DTI acquisition.
• To improve the accuracy and reduce blurring and noise in DTI of ex vivo brain specimens.

Main Methods:

• A model-based compressed sensing framework was used to reconstruct diffusion tensor fields from undersampled k-space data.
• The proposed method was compared against alternative acceleration techniques (reduced resolution, fewer directions, standard CS, asymmetrical sampling).
• Accuracy was assessed for white matter fiber orientation, fractional anisotropy, and mean diffusivity.

Main Results:

• The model-based CS approach reduced image blurring and noise compared to alternative methods.
• It provided more accurate measurements of fiber orientation, fractional anisotropy, and mean diffusivity.
• Measurement accuracy was comparable to fully sampled DTI with significantly fewer encoding directions and shorter scan times.

Conclusions:

• Model-based compressed sensing is a promising technique for accelerating Diffusion Tensor Imaging.
• This method can enhance resolution, accuracy, and reduce scan times for ex vivo DTI studies.
• It offers a viable solution to overcome the limitations of long acquisition times in high-resolution DTI.