Deep Learning and Atlas-Based MRI Segmentation Enable Longitudinal Characterization of Healthy Mouse Brain.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Biomedical Engineering

Background:

• Accurate brain magnetic resonance image (MRI) segmentation is crucial for longitudinal studies.
• Traditional atlas-based segmentation methods can be computationally intensive.
• Deep learning presents a potential alternative for efficient image analysis.

Purpose of the Study:

• To compare the performance of deep learning against atlas-based segmentation for longitudinal mouse brain MRI.
• To evaluate the speed and resource efficiency of deep learning methods.
• To investigate sex-dependent morphological changes in the aging mouse brain using advanced segmentation techniques.

Main Methods:

• Longitudinal brain MRI dataset from adult C57Bl6/J mice.
• Comparison of deep learning segmentation with an atlas-based approach.
• Analysis of computational time and resource requirements for both methods.

Main Results:

• Deep learning segmentation achieved comparable accuracy to atlas-based methods.
• Deep learning demonstrated significantly faster processing times.
• Both methods facilitated large-scale analysis, revealing sex-dependent brain morphology differences in aging mice.

Conclusions:

• Deep learning is a viable, high-throughput tool for longitudinal neuroimaging analysis.
• The study highlights the importance of considering sex as a biological variable in preclinical brain research.
• Efficient segmentation methods like deep learning can accelerate discoveries in aging brain research.