Towards Tunable Consensus Clustering for Studying Functional Brain Connectivity During Affective Processing.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Brain Imaging Analysis

Background:

• Neuroimaging, especially functional magnetic resonance imaging (fMRI), is crucial for understanding brain architecture.
• Data-driven functional connectivity analyses are increasingly popular but face reliability challenges due to numerous methods and small sample sizes.
• Discrepancies in findings arise from the selection of analysis methods and sample limitations in neuroimaging research.

Purpose of the Study:

• To propose a tunable consensus clustering paradigm to address the method selection and reliability issues in neuroimaging analysis.
• To enhance the robustness of findings derived from complex functional connectivity data.
• To validate a novel approach for integrating results from multiple clustering algorithms.

Main Methods:

• A tunable consensus clustering paradigm was developed, integrating results from multiple clustering algorithms applied to various datasets.
• The method was validated using a complex fMRI experiment focusing on affective processing of music clips.
• The approach involved applying three distinct analysis methods and synthesizing their clustering outcomes.

Main Results:

• The consensus clustering method identified intrinsic spatial patterns of coherent neuroactivity in brain structures related to visual, reward, and auditory processing during affective tasks.
• Comparisons showed the proposed paradigm offers significant advantages over traditional single clustering algorithms.
• The method successfully evidenced robust connectivity patterns in complex neuroimaging data with diverse stimuli and affective evaluations.

Conclusions:

• The tunable consensus clustering paradigm effectively overcomes limitations of single clustering methods in neuroimaging.
• This approach enhances the reliability and robustness of functional connectivity findings from fMRI data.
• The developed method provides a valuable tool for analyzing complex neuroimaging datasets, with an implementation available in the R package "UNCLES".