A ResNet mini architecture for brain age prediction.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Imaging

Background:

• Human brain undergoes age-related structural and functional changes.
• Brain age prediction is crucial for assessing neurodevelopment, aging, and disease diagnosis.
• Existing machine learning methods for brain age prediction can be resource-intensive.

Purpose of the Study:

• To develop an efficient deep learning architecture for brain age prediction.
• To reduce memory resource consumption in brain age prediction models.
• To evaluate the performance of the proposed architecture against established networks.

Main Methods:

• A modified 10-layer deep residual neural network (ResNet mini architecture) was developed.
• The ResNet mini architecture was trained and validated using two datasets: OpenNeuro #ds000228 and an Alzheimer's MRI dataset.
• Performance comparison was conducted against other popular neural network architectures.

Main Results:

• The ResNet mini architecture achieved high accuracy in brain age prediction.
• The proposed architecture demonstrated robustness and generality across different datasets.
• The model requires fewer parameters compared to other networks, indicating reduced memory consumption.

Conclusions:

• The ResNet mini architecture is a viable and efficient tool for brain age prediction.
• This compact model offers a promising approach for analyzing brain development and aging.
• The findings support the use of this architecture for neurodevelopment and disease-related research.