Topologically Optimized Intrinsic Brain Networks.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Network Science

Background:

• Accurate estimation of subject-level brain networks is crucial for understanding brain function.
• Existing group-inference methods use either strong voxel-wise constraints or no constraints, each with limitations.
• Strong constraints can obscure individual variability and introduce model bias.

Purpose of the Study:

• To introduce a novel method, Topologically Optimized Intrinsic Brain Networks (TOIBN), for improved subject-level brain network estimation.
• To leverage the topological information of group-level networks as a soft constraint.
• To enhance estimation efficacy in challenging conditions like noisy data or small sample sizes while preserving individual differences.

Main Methods:

• Developed the Topologically Optimized Intrinsic Brain Networks (TOIBN) framework.
• Utilized topological properties of group-level networks to create high-level spatial constraints.
• Applied these constraints to infer subject-level brain networks.

Main Results:

• TOIBN subject maps are less noisy and more aligned with group networks compared to unconstrained methods.
• The method preserves subject-specific variability often lost with strict voxel-wise constraints.
• Topological properties derived from TOIBN maps effectively differentiate individuals with schizophrenia from controls in key brain networks.

Conclusions:

• TOIBN offers a balanced approach, combining the benefits of constraint-based methods with greater flexibility.
• The method improves the accuracy and robustness of subject-level brain network estimation.
• TOIBN has potential applications in clinical neuroscience, particularly in identifying neural differences in psychiatric disorders.