Wavelet-based multifractal analysis of fMRI time series

Area of Science:

• Neuroimaging
• Complex Systems Analysis
• Signal Processing

Background:

• Functional magnetic resonance imaging (fMRI) generates complex time series data.
• Analyzing the intricate dynamics of fMRI signals is crucial for understanding brain activity.
• Traditional methods may not fully capture the subtle, scale-dependent characteristics of neural signals.

Purpose of the Study:

• To investigate fMRI time series using a multifractal approach.
• To extract local singularity exponents and characterize the fractal properties of brain activity.
• To develop a method for identifying activated brain areas based on multifractal parameters.

Main Methods:

• Application of the Wavelet Modulus Maxima (WTMM) method to fMRI time series.
• Quantification of singularity exponent spectra using spectral characteristics (maximum, dimension, FWHM).
• Development and validation of a combined multifractal parameter for brain activation detection.

Main Results:

• Distinct ranges of Hölder exponents were observed in activated (≈1) versus non-activated white matter (≈0.5) voxels.
• A decrease in maximum dimension was noted from white matter to gray matter, and further in activated areas.
• Higher full-width-at-half-maximum (FWHM) values were found in activated brain regions.
• The combined multifractal parameter effectively distinguished activated areas.

Conclusions:

• Multifractal analysis provides a sensitive tool for characterizing fMRI time series dynamics.
• The developed method can reliably identify activated brain regions in both hybrid and in vivo fMRI data.
• This approach offers a paradigm-free method for brain activation detection in neuroimaging research.