Distribution-aware estimation of the minimum achievable uncertainty in diffusion-tensor imaging (DTI).

Area of Science:

• Neuroimaging
• Biomedical Engineering
• Medical Physics

Background:

• Diffusion Tensor Imaging (DTI) is crucial for assessing brain tissue microstructure.
• Standard DTI methods use echo-planar imaging, which is susceptible to noise.
• Current DTI analysis often overlooks MRI noise distribution, potentially affecting accuracy.

Purpose of the Study:

• To develop a novel distribution-aware maximum likelihood tensor estimation framework for DTI.
• To enable separate local noise estimation in diffusion-weighted and reference images.
• To provide guidelines for optimizing DTI acquisition protocols based on SNR and diffusion directions.

Main Methods:

• Developed a distribution-aware maximum likelihood tensor estimation framework.
• Implemented separate local noise estimation for DTI images.
• Validated the framework with water phantom acquisitions and human data.
• Utilized Monte Carlo simulations to analyze DTI uncertainty dependencies.

Main Results:

• The proposed framework accurately estimates DTI parameters while accounting for noise.
• Demonstrated feasibility in both phantom and human brain imaging.
• Identified power-law relationships between DTI uncertainty, SNR, and diffusion directions.

Conclusions:

• The novel framework enhances the reliability of DTI analysis by incorporating noise characteristics.
• This approach offers a pathway to more precise DTI protocol design.
• Findings can guide the optimization of DTI for specific clinical and research applications.