A two-step clustering-based pipeline for big dynamic functional network connectivity data.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Data Science

Background:

• Dynamic functional network connectivity (dFNC) analyzes temporal changes in brain network interactions using resting-state functional magnetic resonance imaging (rs-fMRI).
• Clustering methods are commonly used to identify distinct dFNC states.
• The computational expense of clustering large dFNC datasets presents a significant bottleneck.

Purpose of the Study:

• To develop an efficient dFNC analysis pipeline for large-scale neuroimaging datasets.
• To reduce the computational burden associated with dFNC state identification.
• To offer a scalable solution for analyzing dynamic functional connectivity.

Main Methods:

• A novel dFNC analysis pipeline was developed to process large rs-fMRI datasets efficiently.
• The proposed pipeline was validated against the standard clustering approach using a public dataset.
• Performance was evaluated based on speed and clustering quality.

Main Results:

• The new dFNC pipeline demonstrated comparable dFNC state identification to traditional methods.
• Our approach achieved a 27-fold increase in speed compared to the standard method.
• The pipeline effectively identified optimal cluster numbers and improved clustering quality.

Conclusions:

• The developed dFNC pipeline offers a computationally efficient alternative for analyzing large neuroimaging datasets.
• This method significantly reduces processing time without compromising the quality of dFNC state analysis.
• The pipeline facilitates the study of dynamic brain connectivity in large cohorts.