Sparse solution of fiber orientation distribution function by diffusion decomposition.

Area of Science:

• Neuroimaging
• Diffusion MRI
• Computational Neuroscience

Background:

• Accurate fiber orientation estimation is crucial for diffusion tractography.
• Current deconvolution methods using L2 regularization can produce false fibers, reducing result specificity.
• Existing methods struggle with complex fiber crossings and low signal-to-noise ratio (SNR) data.

Purpose of the Study:

• To introduce diffusion decomposition, a novel method for obtaining sparse fiber orientation distribution functions (ODFs).
• To evaluate the performance of diffusion decomposition against established methods using simulation, phantom, and in-vivo data.
• To demonstrate the method's ability to improve angular resolution and resolve crossing fibers.

Main Methods:

• Diffusion decomposition decomposes the diffusion ODF obtained from various diffusion MRI techniques (QBI, DSI, GQI).
• Performance was assessed via simulation, phantom, and in-vivo studies.
• Compared diffusion decomposition against constrained spherical deconvolution and the ball-and-sticks model.

Main Results:

• Diffusion decomposition showed higher accuracy in simulations compared to constrained spherical deconvolution and ball-and-sticks.
• Phantom studies revealed significantly lower angular error for diffusion decomposition at fiber crossings.
• In-vivo results demonstrated consistent fiber orientation resolution across different diffusion sampling schemes and reconstruction methods.

Conclusions:

• Diffusion decomposition effectively improves angular resolution and resolves crossing fibers, even in low SNR and reduced direction datasets.
• The method shows robustness across different diffusion MRI acquisition and reconstruction strategies.
• Diffusion decomposition holds significant potential for advancing human connectome studies and clinical neuroimaging research.