Brain age prediction using combined deep convolutional neural network and multi-layer perceptron algorithms.

Area of Science:

• Neuroimaging
• Artificial Intelligence
• Gerontology

Background:

• Brain age prediction is crucial for understanding neurodegenerative disease onset and prognosis.
• Existing models often lack the precision needed for nuanced clinical applications.
• Biological sex is a key demographic factor influencing brain structure and aging.

Purpose of the Study:

• To develop and validate a deep learning algorithm for enhanced brain age prediction.
• To investigate the impact of integrating biological sex information on prediction accuracy.
• To assess the algorithm's utility in distinguishing between different neurodegenerative disease groups.

Main Methods:

• Utilized a dataset of 3004 healthy subjects (aged 18+).
• Employed a convolutional neural network (CNN) for analyzing T1-weighted structural brain images.
• Integrated categorical sex information using a multi-layer perceptron (MLP) to create a hybrid CNN-MLP model.
• Compared the hybrid model against a CNN-only model and the established brainageR algorithm.

Main Results:

• The hybrid CNN-MLP algorithm demonstrated superior age prediction accuracy compared to the CNN-only model and brainageR.
• Integration of sex information significantly enhanced the predictive performance of the deep learning model.
• The algorithm successfully identified differences in brain age gaps between mild cognitive impairment and Alzheimer's disease groups.

Conclusions:

• The proposed hybrid deep learning approach offers enhanced precision in brain age prediction.
• Incorporating biological sex information is a valuable strategy for improving neuroimaging-based age estimation.
• This algorithm shows significant potential for clinical applications in diagnosing and monitoring neurodegenerative diseases.