Spectral normative modeling of brain structure.

Area of Science:

• Neuroscience
• Medical Imaging
• Computational Biology

Background:

• Normative modeling in neuroscience characterizes brain variations using reference ranges (brain charts).
• Current models are limited by coarse spatial scales and fixed brain atlases, hindering specificity and adaptability.
• Computational constraints restrict the spatial resolution of existing normative modeling techniques.

Purpose of the Study:

• To introduce spectral normative modeling (SNM) for high-resolution, adaptable brain charting.
• To overcome computational limitations and fixed parcellation dependencies of conventional methods.
• To enable precise characterization of interindividual brain variations across all spatial scales.

Main Methods:

• Leveraging brain eigenmodes for efficient spatial reconstruction in normative modeling.
• Developing SNM to generate normative ranges for arbitrary regions of interest.
• Benchmarking SNM against conventional normative modeling approaches for speed and accuracy.

Main Results:

• SNM achieves a 98.3% speedup in computing accurate normative ranges compared to conventional methods.
• The method provides accurate normative ranges across spatial scales, from millimeters to the whole brain.
• Demonstrated utility in elucidating high-resolution cortical atrophy and characterizing Alzheimer's disease neurodegeneration.

Conclusions:

• SNM offers a computationally efficient and spatially precise approach to normative brain modeling.
• This technique facilitates the creation of adaptable, high-resolution brain charts.
• SNM has the potential to advance individualized precision medicine by enabling detailed analysis of brain phenotypes and diseases.