Joint Multi-Fiber NODDI Parameter Estimation and Tractography Using the Unscented Information Filter.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Biophysics

Background:

• Accurate white matter tractography is crucial for understanding brain connectivity.
• Estimating microstructural tissue parameters is vital for neuroscience applications.
• Previous methods estimated neurite orientation dispersion diffusion imaging (NODDI) parameters independently per voxel.

Purpose of the Study:

• To integrate the NODDI model into a tractography framework for consistent fiber parameter estimation.
• To extend the framework for estimating NODDI parameters in regions with crossing fibers.
• To improve the accuracy and robustness of fiber tract estimation and microstructural analysis.

Main Methods:

• Developed a joint fiber model estimation and tractography algorithm using the NODDI model.
• Integrated NODDI into a tractography framework for the first time.
• Employed the unscented information filter (UIF) for parameter estimation and tractography, offering advantages over the unscented Kalman filter (UKF).
• Extended the framework to handle two crossing fibers and estimate parameters for each bundle separately.

Main Results:

• Demonstrated consistent and smooth estimation of fiber orientation dispersion along tracts.
• Successfully estimated intracellular and extracellular volume fractions from diffusion signals.
• Showcased the algorithm's ability to trace through crossing fiber regions in in-vivo human brain data.
• Provided estimates of parameter confidence via covariance and Fisher-information matrices.

Conclusions:

• The proposed joint NODDI-tractography framework enables accurate tracing of white matter bundles through crossings.
• The method consistently estimates microstructural parameters, including orientation dispersion, along fiber tracts.
• The approach offers significant computational and numerical advantages, enhancing tractography and microstructural analysis in neuroscience.