Using temporal ICA to selectively remove global noise while preserving global signal in functional MRI data

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Biophysics

Background:

• Functional Magnetic Resonance Imaging (fMRI) is crucial for studying brain activity and connectivity.
• fMRI data are contaminated by structured temporal noise from motion, physiology, and equipment.
• Existing spatial Independent Component Analysis (ICA) methods effectively remove spatially specific noise but not global noise.

Purpose of the Study:

• To introduce and validate temporal ICA for removing global structured noise from fMRI data.
• To compare temporal ICA with global signal regression.
• To demonstrate that temporal ICA preserves global neural signals while eliminating physiological noise.

Main Methods:

• Application of temporal ICA to task-based and resting-state fMRI data.
• Comparison of fMRI data processed with temporal ICA versus global signal regression.
• Analysis of global positive and network-specific negative biases.

Main Results:

• Temporal ICA successfully segregates and removes global structured noise.
• Temporal ICA retains the global neural signal in fMRI data.
• Temporal ICA cleanup eliminates global positive biases from physiological noise without introducing negative biases seen with global signal regression.

Conclusions:

• Temporal ICA offers a "best of both worlds" solution to the global signal and noise problem in fMRI.
• This method allows for the selective removal of global noise while preserving neural signals.
• Temporal ICA unlocks new neurobiological insights from fMRI data.