Fiber connectivity integrated brain activation detection.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Medical Image Analysis

Background:

• Functional magnetic resonance imaging (fMRI) for brain activation inference is limited by high noise levels.
• Neural pathways facilitate functional integration, suggesting anatomical connectivity can refine activation detection.

Purpose of the Study:

• To develop a novel method for brain activation detection by integrating anatomical connectivity into fMRI analysis.
• To improve the sensitivity of detecting task-related brain activation and group differences, particularly in noisy data.

Main Methods:

• Formulated activation detection as a probabilistic graph-based segmentation problem.
• Utilized fiber networks from diffusion MRI (dMRI) as a prior for anatomical connectivity.
• Incorporated a data-driven scheme to refine connectivity priors based on task relevance and tractography limitations.

Main Results:

• Demonstrated significantly increased sensitivity in detecting task activation in healthy controls compared to univariate techniques.
• Showcased the model's ability to detect significant group activation differences between schizophrenia patients and controls.
• Highlighted the mitigation of noise confounds in fMRI analysis through anatomical prior integration.

Conclusions:

• Integrating anatomical connectivity derived from dMRI enhances the detection of brain activation from fMRI data.
• The proposed method improves sensitivity and reveals subtle group differences in brain activity, offering a valuable tool for clinical neuroscience research.
• This approach holds promise for advancing the analysis of brain function in various neurological and psychiatric conditions.