Measuring Brain Connectivity via Shape Analysis of fMRI Time Courses and Spectra

Area of Science:

• Neuroimaging
• Data Analysis
• Computational Neuroscience

Background:

• Functional magnetic resonance imaging (fMRI) generates complex time-series data.
• Analyzing fMRI signals requires robust methods for time course and spectral alignment.
• Understanding brain connectivity is crucial for diagnosing neurological and psychiatric disorders.

Purpose of the Study:

• To develop a shape matching approach for aligning fMRI time courses and their spectral properties.
• To enable elastic alignment of both amplitude and phase in fMRI signals.
• To enhance the analysis of brain connectivity, particularly for group comparisons.

Main Methods:

• Utilized concepts from differential geometry and functional data analysis.
• Defined a functional representation for fMRI signals.
• Employed a reparameterization invariant Riemannian metric for elastic alignment of time courses and power spectral densities.

Main Results:

• Demonstrated significant increases in pairwise node-to-node correlations and coherences post-alignment.
• Validated the method's effectiveness in enhancing signal comparability.
• Successfully applied the alignment technique to differentiate connectivity patterns between major depression patients and healthy controls.

Conclusions:

• The proposed shape matching method effectively aligns fMRI time courses and spectral data.
• This alignment improves the sensitivity of connectivity analyses.
• The approach shows promise for identifying neuroimaging biomarkers in psychiatric conditions like major depression.