Higher-order sonification of the human brain.

Area of Science:

• Neuroimaging and Data Sonification
• Computational Neuroscience
• Data Visualization and Accessibility

Background:

• Sonification, converting data to sound, is valuable for science communication and accessibility, especially for visually impaired individuals.
• Current sonification methods struggle with multi-dimensional data, limiting their application.
• Structural magnetic resonance imaging (MRI) data offers rich, multi-dimensional insights into brain structure.

Purpose of the Study:

• To develop and demonstrate a novel sonification technique for multi-dimensional structural MRI data.
• To encode higher-order statistical measures of brain structure into sound for exploratory analysis.
• To assess the fidelity of sonification in representing complex, multi-dimensional neuroimaging data.

Main Methods:

• Analyzed structural MRI data using Fourier space statistical measures: power spectrum and bispectrum.
• Quantified spatial correlations in 3D voxel intensity distributions to derive reduced bispectra.
• Generated audio renderings of a specific reduced-bispectrum configuration (Q019036) from the OASIS-3 dataset for different age groups.

Main Results:

• The developed sonification method successfully encoded multi-dimensional brain MRI data into sound.
• Information loss during reconstruction of bispectra from the sonified signal was minimal, particularly for age-sensitive configurations.
• Audio renderings provided a novel way to explore data patterns related to brain aging across age groups.

Conclusions:

• The presented framework offers a generalizable method for sonifying multi-dimensional data, expanding sonification's utility.
• This approach has the potential to improve scientific exploration and data accessibility for a wider audience.
• Further research is needed for systematic validation of statistical inference and perceptual effectiveness.