Related Experiment Video
Updated: May 26, 2026
16:23
Fiber Connections of the Supplementary Motor Area Revisited: Methodology of Fiber Dissection, DTI, and Three Dimensional Documentation
Published on: May 23, 2017
DeepFixel: Crossing white matter fiber identification through spherical convolutional neural networks
Adam M Saunders1, Lucas W Remedios2, Elyssa M McMaster1
1Department of Electrical and Computer Engineering, Vanderbilt University, Nashville, TN, USA.
Summary
DeepFixel, a novel deep learning method, accurately separates complex crossing white matter fibers in brain imaging. This computationally efficient approach improves structural connectivity analysis and tractography accuracy.
Area of Science:
- Neuroimaging
- Computational Neuroscience
- Biomedical Engineering
Background:
- Diffusion-weighted magnetic resonance imaging (dMRI) enables brain structural connectivity modeling.
- Fiber Orientation Distribution Functions (ODFs) represent white matter fiber bundles but are complicated by crossing fibers.
- Existing methods for separating crossing fibers are computationally intensive or less accurate.
Purpose of the Study:
- To introduce DeepFixel, a spherical convolutional neural network, as an efficient approximation for separating crossing white matter fibers.
- To improve the accuracy and computational efficiency of fiber ODF analysis in neuroimaging.
Main Methods:
- DeepFixel models fiber ODFs using a spherical mesh with high angular resolution.
- It employs a spherical convolutional neural network to approximate a computationally infeasible nonlinear optimization.
- Validation involved comparison against nonlinear optimization and a fixel-based separation algorithm.
Main Results:
- DeepFixel achieved a median angular correlation coefficient of 0.973, closely matching nonlinear optimization (1.00).
- It demonstrated superior computational efficiency (0.32 ms/voxel) compared to nonlinear optimization.
- DeepFixel successfully disentangled fibers with smaller angular separations and volume fractions than fixel-based methods.
Conclusions:
- DeepFixel offers a computationally efficient and accurate solution for separating crossing white matter fibers.
- Its spherical mesh representation enhances the disentanglement of complex fiber architectures.
- This method holds promise for advancing neuroimaging analysis and tractography accuracy.
