Deep Learning Models Unveiled Functional Difference Between Cortical Gyri and Sulci.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Brain Imaging

Background:

• The functional differences between cortical gyral (convex) and sulcal (concave) regions remain largely unexplored.
• Neuroimaging reveals distinct structural connectivity profiles in gyri and sulci, suggesting potential functional divergence.

Purpose of the Study:

• To investigate and confirm functional role differences between cortical gyri and sulci.
• To apply deep learning, specifically convolutional neural networks (CNNs), to fMRI data for differentiating gyral and sulcal activity.

Main Methods:

• Utilized a convolutional neural network (CNN) model to analyze functional magnetic resonance imaging (fMRI) signals from gyral and sulcal regions.
• Applied the CNN model to Human Connectome Project (HCP) and macaque brain fMRI datasets (task and resting-state).

Main Results:

• Achieved high classification accuracy (83-90% for HCP, 78-86% for macaque) in distinguishing gyral and sulcal fMRI signals.
• Identified interpretable features learned by CNNs, revealing functional distinctions between gyri and sulci.

Conclusions:

• Cortical gyri function as global integration centers with low-frequency fMRI signal components.
• Cortical sulci act as local processing units characterized by complex, high-frequency fMRI signal components.
• Deep learning effectively deciphers functional differences in brain regions based on fMRI data.