Scale-dependent brain age with higher-order statistics from structural magnetic resonance imaging

Area of Science:

• Neuroscience
• Cosmology
• Medical Imaging

Background:

• Chronological age inference from magnetic resonance imaging (MRI) is crucial for early neurodegenerative disease detection.
• Existing methods may lack physiological interpretability and scale-specific insights into brain aging.

Purpose of the Study:

• To develop and validate a novel method for chronological age prediction from brain MRI data.
• To leverage cosmological techniques for enhanced analysis of brain structure evolution.
• To provide physiologically interpretable insights into age-related brain changes at different anatomical scales.

Main Methods:

• Applied a cosmological technique using higher-order summary statistics with multivariate two- and three-point analyses in 3D Fourier space.
• Utilized MRI data from the OASIS-3 database (869 sessions) for method development.
• Validated the method on the Cam-CAN dataset and simulated MRI images from a Latent Diffusion model.

Main Results:

• Achieved a Mean Absolute Error (MAE) of 3.1 years for chronological age prediction using OASIS-3 data.
• Demonstrated scale-dependent brain aging patterns: structure loss on larger scales (ventricular expansion) and increased structure on smaller scales (cortical thickness).
• Validation on Cam-CAN dataset yielded an MAE of ~5.9 years (age 18-88).

Conclusions:

• The interdisciplinary approach, bridging cosmology and neuroscience, offers a powerful tool for understanding brain aging.
• The method provides physiologically interpretable, scale-specific information about brain structure evolution.
• Individual age prediction uncertainty highlights the influence of genetic or lifestyle factors beyond sample variance.