Canonical cerebellar graph wavelets and their application to FMRI activation mapping

Area of Science:

• Neuroimaging
• Statistical analysis
• Brain mapping

Background:

• Wavelet-based statistical parametric mapping (WSPM) integrates wavelet processing with voxel-wise statistical testing for fMRI.
• Previous WSPM utilized graph wavelets specific to individual gray-matter structure, outperforming classical Euclidean grid wavelets.
• Analysis on subject-invariant graphs necessitates canonical graph wavelets in normalized brain space.

Purpose of the Study:

• To introduce a method for designing canonical cerebellar graph wavelets in normalized brain space.
• To evaluate the performance of these cerebellar graph wavelets in fMRI activation mapping.

Main Methods:

• A fixed template graph of the cerebellum was created using the SUIT cerebellar template.
• Canonical cerebellar graph wavelets were constructed based on this template graph.
• These wavelets were applied to analyze synthetic and real fMRI data.

Main Results:

• WSPM with cerebellar graph wavelets demonstrated superior type-I error control compared to classical SPM.
• The method showed empirically higher sensitivity in detecting activity in real fMRI data.
• The approach has the potential to identify subtle patterns of cerebellar activity.

Conclusions:

• Canonical cerebellar graph wavelets provide improved statistical power and accuracy in fMRI.
• This method advances brain activity detection, particularly within the cerebellum.
• WSPM with tailored graph wavelets offers a more sensitive approach to neuroimaging analysis.