Anti-Fragmentation of Resting-State Functional Magnetic Resonance Imaging Connectivity Networks with Node-Wise...

Area of Science:

• Neuroimaging
• Network Science
• Computational Neuroscience

Background:

• Functional magnetic resonance imaging (fMRI) is used to construct functional connectivity networks.
• Network construction typically involves thresholding correlation matrices of nodal time courses.
• Hard thresholding, a common method, can lead to fragmented networks with disconnected nodes.

Purpose of the Study:

• To compare network characteristics between hard thresholding and node-wise thresholding methods.
• To evaluate the impact of thresholding strategies on network fragmentation and modular organization.
• To assess the robustness of network properties derived from different thresholding approaches.

Main Methods:

• Analysis of publicly available resting-state fMRI data from 123 healthy young subjects.
• Construction of functional connectivity networks using both hard thresholding and node-wise thresholding.
• Comparison of network fragmentation, number of disconnected nodes, and modular organization between the two methods.

Main Results:

• Hard thresholding resulted in networks with a significant number of disconnected nodes and fragmented modular organization.
• Node-wise thresholding produced networks with less fragmentation and more robust, consistent modular structures.
• Node-wise thresholding demonstrated reduced sensitivity to threshold selection, yielding stable network properties.

Conclusions:

• Node-wise thresholding is a superior method for constructing less fragmented functional connectivity networks from fMRI data.
• This approach facilitates a more robust and reliable characterization of brain network properties.
• Node-wise thresholding offers advantages in terms of network consistency and reduced susceptibility to arbitrary threshold choices.