Human brain mapping: A systematic comparison of parcellation methods for the human cerebral cortex

Area of Science:

• Neuroscience
• Brain Imaging
• Network Analysis

Background:

• The macro-connectome represents brain connections as a network, crucial for cognitive tasks.
• Defining network nodes (brain regions) is critical for connectivity analysis.
• Traditional anatomical atlases are being supplemented by newer connectivity-driven parcellation methods.

Purpose of the Study:

• To systematically compare anatomical, connectivity-driven, and random brain parcellation methods.
• To evaluate parcellation accuracy using diverse quantitative metrics and data types.
• To provide guidance on selecting appropriate parcellation techniques and resolutions.

Main Methods:

• Utilized resting-state functional MRI data from the Human Connectome Project.
• Evaluated 10 subject-level and 24 groupwise parcellation methods at multiple resolutions.
• Assessed parcellations based on reproducibility, connectivity fidelity, agreement with other brain maps, and network analysis.

Main Results:

• No single parcellation method demonstrated superiority across all evaluation criteria.
• Connectivity-driven methods showed promise in delineating functionally coherent regions.
• Strengths and weaknesses of various methods varied depending on resolution and application.

Conclusions:

• The optimal brain parcellation strategy depends on the specific research question and data.
• Further research is needed to refine parcellation techniques for comprehensive brain network analysis.
• This study offers a framework for selecting appropriate parcellation methods in neuroscience research.