Depthgram: Visualizing outliers in high-dimensional functional data with application to fMRI data exploration.

Area of Science:

• Neuroscience
• Medical Imaging
• Data Visualization

Background:

• Functional magnetic resonance imaging (fMRI) measures brain activity via blood flow changes.
• Task fMRI (tfMRI) and resting-state fMRI (rsfMRI) provide insights into brain organization and baseline activity.
• High-resolution fMRI generates vast amounts of longitudinal data for each individual.

Purpose of the Study:

• To develop novel visualization techniques for high-dimensional fMRI data.
• To create computationally efficient 2-dimensional representations of fMRI data.
• To elucidate sample composition, outlier presence, and individual variability in fMRI datasets.

Main Methods:

• Utilized depth-based notions for data representation.
• Developed computationally efficient methods for dimensionality reduction.
• Applied techniques to motor and language tfMRI experiments.

Main Results:

• Demonstrated the ability of visualization techniques to represent complex fMRI data.
• Showcased the elucidation of sample composition and outlier detection.
• Provided efficient 2-dim representations of high-dimensional functional brain data.

Conclusions:

• The proposed visualization techniques are crucial for inferential approaches in neuroscience.
• These methods facilitate the identification of neuroscientific patterns across individuals, tasks, and brain regions.
• Effective visualization is key to understanding complex fMRI data and brain function.