Decoding of columnar-level organization across cortical depth using BOLD- and CBV-fMRI at 7 T

Area of Science:

• Neuroimaging
• Cognitive Neuroscience
• Human Brain Mapping

Background:

• Functional magnetic resonance imaging (fMRI) measures hemodynamic responses, which are spatially limited by macrovascular contributions.
• Multivariate pattern analysis (MVPA) leverages multi-voxel information to retrieve fine-grained spatial patterns from neuroimaging data.
• Imaging cortical structures at the scale of columns and layers is challenging due to fMRI signal limitations.

Purpose of the Study:

• To investigate the spatial specificity of signals used by MVPA in high-resolution fMRI.
• To assess the impact of macrovascular contributions on MVPA decoding across cortical depth.
• To compare different fMRI acquisition techniques for their ability to resolve laminar information.

Main Methods:

• Acquisition of 7 Tesla (7 T) fMRI data in human primary visual cortex (V1) using gradient echo-based BOLD (GE-BOLD), spin echo-based BOLD (SE-BOLD), and vascular-space-occupancy (VASO) techniques.
• Measurement of ocular dominance columns (ODCs) to serve as a known fine-grained spatial pattern.
• Decoding of eye-of-origin information from fMRI signals across cortical layers.

Main Results:

• Ocularity information was successfully decoded using all tested fMRI acquisition techniques (GE-BOLD, SE-BOLD, VASO).
• Laminar profiles indicated that macrovascular contributions are present in all methods, limiting specificity throughout cortical depth.
• The spatial specificity of MVPA is constrained by inherent limitations in fMRI signal localization.

Conclusions:

• MVPA is a powerful tool for exploring mesoscopic circuitry in the human cerebral cortex.
• All current fMRI acquisition methods are susceptible to macrovascular signal contamination.
• Further research must carefully account for macrovascular effects to improve the spatial specificity of MVPA-based analyses.