Resting state fMRI-guided fiber clustering: methods and applications

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Biomedical Engineering

Background:

• Tract-based analysis of white matter integrity and brain connectivity modeling relies on clustering streamline fibers.
• Existing fiber clustering methods primarily use structural MRI and DTI data, neglecting functional information.

Purpose of the Study:

• To propose and validate a novel multimodal approach for clustering streamline fibers by integrating resting-state fMRI (rsfMRI) and DTI data.
• To utilize functional coherence derived from rsfMRI to guide the clustering of DTI-derived streamline fibers into functionally meaningful bundles.

Main Methods:

• Combined rsfMRI and DTI data for fiber clustering.
• Defined functional coherence based on rsfMRI time series correlations between streamline fibers.
• Employed the Affinity Propagation (AP) algorithm for clustering DTI-derived streamline fibers.
• Validated the methodology using corpus callosum (CC) fibers, task-based fMRI, reproducibility studies, and comparisons with other methods.

Main Results:

• The proposed rsfMRI-guided fiber clustering method yields functionally homogeneous bundles.
• Clustered bundles demonstrate reasonable consistency across individuals and populations.
• Results suggest a strong link between structural connectivity and brain function.
• The framework was successfully applied to a multimodal rsfMRI/DTI dataset of schizophrenia (SZ) with reproducible findings.

Conclusions:

• The multimodal approach effectively integrates structural and functional neuroimaging data for advanced white matter tractography.
• Functional coherence is a valuable criterion for guiding fiber clustering, leading to more meaningful bundles.
• This method holds promise for understanding brain connectivity alterations in neurological and psychiatric disorders like schizophrenia.