On the relation between brain images and brain neural networks

Area of Science:

• Neuroscience
• Computational Neuroscience
• Biophysics

Background:

• Brain imaging techniques like Positron Emission Tomography (PET) and functional Magnetic Resonance Imaging (fMRI) provide insights into brain function.
• Understanding the precise relationship between observed imaging signals and underlying neural activity remains a challenge.

Purpose of the Study:

• To analyze the relationship between brain imaging data (PET, fMRI) and neural activity.
• To develop and validate new structural equation models (SEMs) for brain imaging data analysis.
• To explore the connection between these models and neural coding principles.

Main Methods:

• Utilized recent findings on averaged and synchronized activity in coupled neural networks.
• Employed a simplified model of neural activity-induced blood flow.
• Derived novel SEMs, incorporating neuronal activity as hidden variables.
• Analyzed covariance structural equation modeling of brain imaging data.

Main Results:

• Specified conditions for coupled neural systems leading to SEMs.
• Established a link between derived models and potential neural codes.
• Developed a new SEM framework where all neuronal activity is represented by hidden variables.
• Discussed the transferability of these findings to electroencephalography (EEG) and magnetoencephalography (MEG) data.

Conclusions:

• The study provides a theoretical framework linking brain imaging signals to neural activity through advanced statistical modeling.
• The developed SEMs offer new tools for analyzing complex brain data and understanding neural coding.
• The findings have implications for interpreting data from various neuroimaging modalities, including EEG and MEG.