Brain structure ages-A new biomarker for multi-disease classification.

Area of Science:

• Neuroimaging
• Artificial Intelligence
• Neurology

Background:

• Chronological age is a key factor in brain anatomy and aging.
• Deviations from normal brain aging trajectories may indicate neurological diseases.
• Current brain age gap metrics are insufficient for detailed disease classification.

Purpose of the Study:

• To extend global brain age estimation to individual brain structure ages using structural MRI.
• To develop a novel biomarker for anomaly detection and differential diagnosis of neurological conditions.
• To enhance the clinical utility of neuroimaging in understanding brain aging and disease.

Main Methods:

• Utilized an ensemble of deep learning models to generate a 3D voxel-wise aging map from structural MRI.
• Employed 3D segmentation masks to derive specific brain structure ages.
• Validated the approach for population-level anomaly detection and subject-level disease classification.

Main Results:

• The proposed method for estimating brain structure ages outperforms existing state-of-the-art approaches in anomaly detection.
• Brain structure age deviations provide a powerful feature for multi-disease classification and differential diagnosis.
• Visualizations of individual brain structure age deviations offer clinical insights into brain abnormalities.

Conclusions:

• Brain structure age estimation offers a more informative biomarker than global brain age for neurological disease analysis.
• This approach enhances anomaly detection at the population level and differential diagnosis at the subject level.
• The developed biomarker has significant potential to aid clinicians in medical contexts by visualizing brain abnormalities.