A prediction model for cognitive performance in health ageing using diffusion tensor imaging with graph theory.

Area of Science:

• Neuroimaging
• Graph Theory
• Machine Learning

Background:

• Cognitive decline is a growing concern in aging populations.
• Understanding brain structural networks is crucial for assessing cognitive function.
• Diffusion tensor imaging (DTI) offers insights into brain connectivity.

Purpose of the Study:

• To construct brain structural networks using DTI in healthy elderly subjects.
• To identify network characteristics correlating with cognitive performance.
• To develop machine learning models for predicting cognitive abilities.

Main Methods:

• Diffusion tensor imaging (DTI) was used to create brain structural networks.
• Graph theory analysis derived topological properties from connection matrices.
• Correlation analysis linked network properties with the Cognitive Abilities Screening Instrument (CASI).
• Machine learning algorithms, including linear regression and Gaussian processes, were employed for prediction.

Main Results:

• Significant brain network characteristics were identified and correlated with cognitive performance.
• Linear regression and Gaussian processes models demonstrated high predictive accuracy (MAE 5.8120 and 6.25).
• DTI-derived network properties show potential as biomarkers for cognitive health.

Conclusions:

• Brain structural network properties derived from DTI can predict cognitive performance in the elderly.
• These network characteristics may serve as biomarkers for brain degeneration and early MCI diagnosis.
• Machine learning models effectively utilize DTI-derived network features for cognitive assessment.