Characterizing fiber directional uncertainty in diffusion tensor MRI.

Area of Science:

• Neuroimaging
• Biomedical Engineering
• Diffusion Tensor Imaging

Background:

• Image noise in diffusion tensor MRI (DT-MRI) introduces errors in tensor measurements and fiber orientation estimation.
• The "cone of uncertainty" (CU) is a circular model used to represent variability in DT-MRI parameters and fiber tractography.
• The circular assumption of the CU has not been empirically validated.

Purpose of the Study:

• To investigate the shape and statistical properties of the cone of uncertainty in DT-MRI.
• To determine if the CU is indeed circular or if it exhibits other characteristics.
• To provide a more accurate model for error quantification in DT-MRI analysis.

Main Methods:

• Utilized bootstrap analysis on DT-MRI data.
• Employed theoretical arguments to model the cone of uncertainty.
• Quantified the deviation of the CU's major axis from the tensor's second eigenvector.

Main Results:

• Demonstrated that the cone of uncertainty is predominantly elliptical and multivariate normal in white matter voxels under typical conditions.
• Established the relationship between the cone angle, signal-to-noise ratio (SNR), and eigenvalue contrast.
• Showed that the major and minor axes of the elliptical CU align with the second and third eigenvectors of the tensor, respectively, under ideal conditions.

Conclusions:

• The elliptical CU model offers a more accurate representation of orientation uncertainty in DT-MRI compared to the circular model.
• This finding has significant implications for improving the realism of error information in fiber-tracking algorithms.
• The results provide a quantitative basis for optimizing DT-MRI acquisition protocols.