MRI-based whole-brain elastography and volumetric measurements to predict brain age

Area of Science:

• Neuroimaging
• Biophysics
• Machine Learning

Background:

• Brain age estimation from neuroimaging aids in assessing neurodevelopmental and neurodegenerative conditions.
• Magnetic resonance elastography (MRE) measures brain mechanical properties (stiffness, damping ratio), which change with age and disease.
• These properties correlate with cognitive function and neurodegeneration.

Purpose of the Study:

• To develop a machine learning framework for predicting chronological age from brain mechanical property maps.
• To assess the utility of MRE-derived metrics in brain age modeling.
• To investigate multimodal approaches combining mechanical properties and volumetric data.

Main Methods:

• Utilized 3D convolutional neural networks and traditional statistical models.
• Integrated whole-brain measurements of stiffness, damping ratio, and volume.
• Evaluated model performance on held-out subject data.

Main Results:

• A multimodal approach combining stiffness and volume provided the most accurate age predictions.
• Model interpretation identified key brain regions contributing to age estimation across modalities.
• MRE measurements offer complementary information to traditional brain age models.

Conclusions:

• MRE-derived brain mechanical properties are valuable additions to brain age prediction frameworks.
• Multimodal integration of MRE data with volumetric information enhances prediction accuracy.
• This approach holds potential for improved detection of neuropathology-related brain integrity changes.