neurolib

neurolib implements whole-brain neural mass models to simulate and analyze mesoscopic brain dynamics and to relate simulated activity to empirical fMRI BOLD signals.

Key Features:

• Neural Mass Models: Implements a suite of neural mass models that represent the average activity of neuronal populations within distinct brain regions.
• Biologically Informed Connectivity: Constructs whole-brain network models using structural connectivity (the connectome) to define interregional coupling.
• Data Integration and Model Setup: Supports integration of structural and functional datasets and configuration of model parameters for whole-brain model setup.
• Simulation and Output Management: Simulates brain activity across the entire network and organizes simulation outputs for subsequent analysis.
• BOLD Signal Simulation: Converts simulated neural activity into BOLD (Blood Oxygen Level Dependent) signals for comparison with empirical fMRI data.
• Extensive Model Analysis: Provides a parameter exploration module for systematic analysis of model behavior across parameter spaces.
• Model Optimization: Includes an optimization module that employs evolutionary algorithms to fit models to multimodal empirical data.
• Extensibility: Allows implementation and integration of custom neural mass models to extend the framework.
• Python Implementation: Implemented in Python.

Scientific Applications:

• Structure–Function Relationship: Investigating relationships between structural connectivity and emergent functional dynamics in brain networks.
• In-silico Model Optimization: Performing in-silico optimization and calibration of whole-brain models against multimodal empirical data.
• Cognitive Neuroscience: Exploring how different neural dynamics contribute to cognitive processes.
• Neurological Disorders: Studying alterations in whole-brain dynamics relevant to neurological disorders.

Methodology:

Uses neural mass models with connectome-based interregional coupling, integrates structural and functional datasets, simulates whole-brain activity, converts simulated activity to BOLD signals, performs parameter exploration, and applies evolutionary-algorithm optimization to fit models to multimodal empirical data.