Assessing diffusion kurtosis tensor estimation methods using a digital brain phantom derived from human connectome...

Area of Science:

• Medical Imaging
• Neuroimaging
• Biophysics

Background:

• Diffusion kurtosis imaging (DKI) is an advanced diffusion-weighted MRI technique gaining prominence.
• Accurate processing algorithms are crucial for reliable DKI data interpretation.
• Quantitative comparison of existing algorithms is needed to guide clinical adoption.

Purpose of the Study:

• To quantitatively compare the performance and accuracy of four DKI processing algorithms.
• To develop a digital DKI brain phantom for algorithm evaluation.
• To assess algorithm performance across different regions and noise levels.

Main Methods:

• A digital DKI brain phantom was generated using Human Connectome Project data.
• Four DKI processing algorithms were evaluated using Monte Carlo Rician noise simulations.
• Performance was assessed based on mean squared error, squared bias, and variance.

Main Results:

• Algorithm performance demonstrated significant region-dependency and variability across diffusion metrics.
• A crossover point for variance and squared bias error was observed between SNR levels of 30 and 40.
• Spatial variations in algorithm performance necessitate the use of error maps.

Conclusions:

• The developed framework allows for quantitative comparison of DKI algorithms using ground truth data.
• Bias-plus-variance decomposition offers a more comprehensive evaluation than MSE alone.
• Future refinements in acquisition and processing will enhance DKI accuracy and efficiency for clinical use.