Visualizing temporal brain-state changes for fMRI using t-distributed stochastic neighbor embedding.

Area of Science:

• Neuroimaging
• Data Visualization
• Machine Learning

Background:

• Functional magnetic resonance imaging (fMRI) is crucial for brain dynamics but visualizing its temporal data is challenging.
• Complexity of 4D fMRI data hinders direct interpretation of temporal aspects.

Purpose of the Study:

• To present a t-distributed stochastic neighbor embedding (t-SNE) postprocessing technique for visualizing temporal information in 4D fMRI data.
• To demonstrate the utility of t-SNE in detecting brain meta-state changes during fMRI scans.
• To enable quantification of hemodynamic delay using detected brain-state changes.

Main Methods:

• Application of t-distributed stochastic neighbor embedding (t-SNE) for postprocessing 4D fMRI data.
• Utilizing t-SNE for visualization of temporal dynamics in both task and resting-state fMRI.
• Comparing detected brain-state changes with experimental paradigms to quantify hemodynamic delay.

Main Results:

• The t-SNE method successfully detects brain-state transitions (task-rest and rest-task) in fMRI data.
• Temporal visualization of brain states is achievable for both task and resting-state fMRI.
• Hemodynamic delay can be quantified by correlating t-SNE detected changes with task paradigms.

Conclusions:

• t-SNE visualization effectively identifies major brain-state changes in fMRI data.
• This technique offers insights into subject engagement and attentiveness during scans.
• The method can serve as a quality control tool for subject performance in fMRI studies.