SNAKE: A modular realistic fMRI data simulator from the space-time domain to k-space and back

Area of Science:

• Neuroimaging
• Biomedical Engineering
• Computational Neuroscience

Background:

• Functional magnetic resonance imaging (fMRI) acquisition involves complex processes from neural activity to data sampling.
• Developing and validating new fMRI acquisition and reconstruction techniques requires realistic data with known ground truth.
• Current simulation tools often lack the end-to-end fidelity needed for advanced fMRI acquisition strategies.

Purpose of the Study:

• To introduce SNAKE, an open-source, Python-based simulator for realistic 3D+time fMRI data.
• To provide a flexible platform for simulating the entire fMRI data acquisition chain.
• To generate reproducible ground truth data for evaluating fMRI reconstruction and analysis methods.

Main Methods:

• SNAKE simulates neurovascular coupling, brain responses, and 3D k-space data acquisition with multiple coils.
• The software allows for the extension of the forward acquisition model to include various noise and artifact sources.
• It supports diverse sampling strategies, including 3D Cartesian and non-Cartesian patterns.

Main Results:

• Demonstrated SNAKE's flexibility and fidelity through three scenarios of increasing complexity.
• Showcased the ability to generate realistic, reproducible ground truth for accelerated 3D fMRI acquisition.
• Enabled exploration of critical parameter influences (e.g., acceleration factor, SNR) on reconstruction and statistical analysis.

Conclusions:

• SNAKE offers a powerful, memory-efficient tool for advancing fMRI acquisition technique research.
• The simulator facilitates reproducible benchmarking of experimental paradigms, acquisition strategies, and reconstruction methods.
• This in-silico approach aids in understanding the interplay of factors affecting downstream fMRI statistical analysis.