Should one go for individual- or group-level brain parcellations? A deep-phenotyping benchmark

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Brain Mapping

Background:

• Brain analysis often requires region-level information, moving beyond voxel-sampled data.
• Subject-specific brain parcellations adapt to individual neural organization, improving accuracy over standard atlases.
• A key challenge is maintaining group-level consistency with adaptable, individualized brain representations.

Purpose of the Study:

• To investigate if the benefits of individualized brain models generalize to new subjects.
• To determine if existing parcellations can be adapted to new individuals for accurate summaries.
• To quantitatively assess activity patterns using adapted parcellations in naturalistic and task-based settings.

Main Methods:

• Adapted a dictionary-learning method to generate subject-specific brain parcellations.
• Applied the adapted method to a deep-phenotyping dataset.
• Evaluated the quantitative patterns of brain activity under different experimental conditions.

Main Results:

• Individualized brain parcellations provide substantial representational benefits.
• These benefits are not solely an artifact of circular analysis but generalize to new individuals.
• The advantages of individualized parcellations exhibit significant variability across different brain systems.

Conclusions:

• Adapted individualized brain parcellations are effective for summarizing brain activity in new subjects.
• The utility of subject-specific brain maps is confirmed, demonstrating their value beyond individual-specific analysis.
• Future research should consider the system-dependent nature of benefits when applying individualized parcellations.