Probabilistic tractography using Lasso bootstrap.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Biophysics

Background:

• Diffusion MRI (dMRI) enables noninvasive imaging of white matter tracts.
• Fiber tracking reconstructs white matter tracts using fiber orientations (FOs) from dMRI.
• Probabilistic tractography addresses uncertainties in FO estimation due to image noise.

Purpose of the Study:

• To develop a probabilistic tractography algorithm for dMRI models incorporating sparsity regularization.
• To address the inapplicability of residual bootstrap to new sparsity-regularized models.
• To introduce Lasso bootstrap tractography (LBT) for improved FO estimation and tract reconstruction.

Main Methods:

• Modeled diffusion signals using a Lasso formulation with a fixed tensor basis and sparsity assumption.
• Generated a distribution of diffusion signals using a modified Lasso bootstrap strategy on Lasso model residuals.
• Estimated FOs from synthesized diffusion signals using an algorithm enforcing spatial consistency.
• Performed fiber tracking using the computed FOs.

Main Results:

• The LBT algorithm was evaluated on simulated and real dMRI data.
• Qualitative and quantitative assessments demonstrated LBT's performance.
• LBT was shown to outperform state-of-the-art tractography algorithms.

Conclusions:

• LBT is a novel probabilistic tractography algorithm suitable for sparsity-regularized dMRI models.
• The method effectively estimates fiber orientations and reconstructs white matter tracts.
• LBT offers superior performance compared to existing state-of-the-art algorithms in dMRI analysis.