Joint Reconstruction of Multiple b-Values and Multiple Directions for Accelerating Diffusion Kurtosis Imaging

Area of Science:

• Medical Imaging
• Diffusion MRI
• Quantitative Imaging

Background:

• Multishot interleaved echo-planar imaging (iEPI) offers higher resolution for diffusion kurtosis imaging (DKI) than single-shot EPI.
• Clinical use of iEPI DKI is limited by its long acquisition times.

Purpose of the Study:

• To develop and validate a novel model-based reconstruction approach for accelerating iEPI DKI acquisition.
• To improve the temporal efficiency of high-resolution DKI while maintaining accuracy.

Main Methods:

• A model-based framework was developed for direct DKI tensor estimation from k-space data.
• The method incorporates an intrinsic DKI signal model as a prior, utilizing total variation (TV) regularization and a physically relevant (PhyR) constraint (mDKI-TV-PhyR).
• Performance was evaluated against conventional methods and variants using simulated and in vivo data with 4-fold undersampling.

Main Results:

• The mDKI-TV-PhyR method demonstrated lower root-mean-square error (RMSE) for all DKI parameters compared to conventional methods.
• Bland-Altman analysis indicated the smallest bias and narrowest limits of agreement for fractional anisotropy (FA) and mean kurtosis (MK).
• mDKI-TV-PhyR produced MK maps without artifacts, showed improved stability, and reduced FA bias compared to mDKI-TV.

Conclusions:

• The proposed mDKI-TV-PhyR method significantly accelerates iEPI DKI acquisition.
• This approach holds substantial promise for clinical applications requiring both high temporal efficiency and spatial resolution.
• The method enhances the feasibility of advanced diffusion imaging techniques in clinical settings.