Measuring depth-dependent timing differences in the cortex - Laminar fMRI acquisition and analysis strategies

Abstract

Laminar fMRI enables non-invasive studying of hemodynamics arising from changes in neuronal activity across depth. This technique also holds the potential to study submillimeter-scale processing and the direction of information flow between cortical layers. Studying timing differences between layers might provide information on feedforward and feedback loops between layers and areas. Because these timing differences are in the order of tens of milliseconds, this requires not only high spatial but also high temporal resolution. Measuring cortical depth-dependent neuronal activity with fMRI is further complicated by fMRI being an indirect measurement of neuronal activation. Temporal differences in the fMRI signal arise from a combination of neuronal and hemodynamic contributions, which both vary across layers. This review aims to provide an overview of the origins of timing differences in the hemodynamic response and acquisition and analysis strategies and challenges for measuring and interpreting timing differences. First, we provide an overview of the origins of layer- dependent timing differences in neuronal activation and the underlying vasculature in the cortex. Then, we discuss various acquisition techniques that are available for laminar fMRI and we review their potential for high spatiotemporal resolution scanning. We compare various BOLD and non-BOLD strategies and show that BOLD-based fMRI is the most suitable for high temporal resolution scanning. However, BOLD-based methods each come with their own challenges in terms of sensitivity and specificity. Lastly, we consider analysis methods that increase the sensitivity and the specificity of laminar fMRI results after acquisition. This includes methods to remove vascular contributions from the signal and for denoising the signal. Together, a deeper understanding of the cortical vasculature and advanced acquisition and analysis techniques may bring us closer to measuring laminar differences in neuronal activation and correct interpretation of laminar fMRI results. This ultimately improves our understanding of information flow across cortical layers.