Unified surface and volumetric projection of physiological imaging data

Abstract

Projection of volumetric data onto the cortical surface is an important precursor to performing surface-based analysis. Numerous projection methods have been reported in the literature, many of which make assumptions which tie them to use with specific modalities, notably blood oxygenation level dependent (BOLD) imaging. This means that they may not be appropriate for use with modalities where subcortical tissue contributes a signal of interest. This work details a new projection that provides a number of generalisations and extensions to existing methods. Namely, it may be used to project arbitrary data without making modality-specific assumptions and can produce unified surface and volumetric representations of data (a concept also known as grayordinates space). When constrained to the same assumptions as existing methods, a comparison using simulation data shows that it produces similar outputs. When these assumptions are relaxed to project simulation data containing both cortical and subcortical signals to and from a unified surface and volume space, substantial and statistically significant differences in recovery of ground truth are observed compared to existing methods.