A novel patch-based procedure for estimating brain age across adulthood

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Gerontology

Background:

• Brain aging involves structural changes detectable via neuroimaging.
• Monitoring brain alterations across the lifespan is crucial for understanding neurological health.
• Identifying deviations from typical aging patterns can inform early detection of cognitive changes.

Purpose of the Study:

• To develop and validate a novel patch-based grading procedure for estimating cortical brain age.
• To assess the reliability and robustness of the proposed method in cognitively intact individuals.
• To provide a tool for understanding brain aging trajectories and potential departures from normal aging.

Main Methods:

• A machine learning algorithm was employed for a patch-based grading procedure.
• The method computed cortical age based on image similarity between test and training cortical labels.
• The framework was trained on 100 cognitively intact individuals (aged 19-61) and tested on 78 others.

Main Results:

• The patch-based framework achieved a high R-squared value of 0.94.
• A low mean absolute error of 1.66 years was reported, indicating high accuracy.
• The method demonstrated reliability and robustness even with a limited training dataset.

Conclusions:

• The proposed patch-based grading framework is a reliable and robust method for estimating brain age from neuroimaging data.
• This approach offers a valuable tool for assessing cortical aging in research and clinical settings.
• The method's accuracy supports its utility in identifying deviations from typical brain aging patterns.