Sparse multivariate autoregressive modeling for mild cognitive impairment classification.

Area of Science:

• Neuroimaging
• Systems Neuroscience
• Biomedical Engineering

Background:

• Functional magnetic resonance imaging (fMRI) is crucial for studying cognitive and perceptual processes.
• Understanding directed functional interactions in brain networks is challenging.
• Sparse network models enhance biological plausibility for genetic and biomedical data.

Purpose of the Study:

• To develop an effective connectivity model for resting-state fMRI data.
• To incorporate sparsity for biologically meaningful interpretations.
• To improve classification of mild cognitive impairment (MCI).

Main Methods:

• Utilized multivariate autoregressive (MAR) modeling to characterize dynamic systems.
• Applied Granger causality to identify effective connectivity.
• Employed forward orthogonal least squares (OLS) regression for sparse MAR model construction.

Main Results:

• Identified key discriminative regions for MCI classification: middle cingulate gyrus, posterior cingulate gyrus, lingual gyrus, and caudate.
• Achieved a classification accuracy of 91.89% for MCI.
• Demonstrated a 5.4% accuracy improvement over non-directional functional connectivity methods.

Conclusions:

• The proposed sparse MAR model effectively captures directed functional interactions in fMRI data.
• This approach enhances the biological interpretability and classification performance for neurological conditions like MCI.
• The findings highlight the potential of Granger causality and sparse modeling in brain network analysis.