Identifying left and right hemispheres using functional connectivity

Area of Science:

• Neuroscience
• Cognitive Science
• Machine Learning

Background:

• The human brain's left and right hemispheres exhibit organizational differences, particularly in language processing.
• Understanding hemispheric specialization is crucial for neuroscience and clinical applications.
• Functional connectivity patterns offer a novel approach to studying brain organization.

Purpose of the Study:

• To determine if supervised learning can classify brain hemispheres (left vs. right) based on functional connectivity data.
• To investigate the influence of handedness on the accuracy of hemispheric classification.
• To explore the potential of using functional connectivity to identify handedness and hemisphere chirality simultaneously.

Main Methods:

• Utilized functional connectivity data from the Human Connectome Project.
• Employed supervised learning algorithms for hemisphere classification.
• Analyzed classification accuracy across participants with varying degrees of handedness (Edinburgh Handedness Inventory [EHI]).

Main Results:

• Achieved high classification accuracies (> .90) for distinguishing left and right hemispheres in right-handed participants (EHI > 0).
• Maintained high, though slightly reduced, accuracies when classifying hemispheres in left-handed participants (EHI ≤ 0).
• Failed to successfully identify handedness alongside hemisphere chirality, but found reduced hemispheric distinctiveness correlated with stronger left-handedness.

Conclusions:

• Supervised learning effectively categorizes human brain hemispheres based on functional connectivity, especially in right-handed individuals.
• Hemispheric organization and distinctiveness are influenced by handedness.
• Findings provide insights into developmental and post-injury hemispheric organization and offer potential for future research.