No abstract

No abstract

Cortical depth-dependent functional magnetic resonance image (fMRI), also known as layer-fMRI, has the potential to capture directional neural information flow of brain computations within and across large-scale cortical brain networks. E.g., layer-fMRI can differentiate feedforward and feedback cortical input in hierarchically organized brain networks. Recent advancements in 3D-EPI sampling approaches and MR contrast generation strategies have allowed proof-of-principle studies showing that layer-fMRI can provide sufficient data quality for capturing laminar changes in functional connectivity. These studies have however not shown how reliable the signal is and how repeatable the respective results are. It is especially unclear whether whole-brain layer-fMRI functional connectivity protocols are widely applicable across common neuroscience-driven analysis approaches. Moreover, there are no established preprocessing fMRI methods that are optimized to work for whole-brain layer-fMRI datasets. In this work, we aimed to serve the field of layer-fMRI and build tools for future routine whole-brain layer-fMRI in application-based neuroscience research. We have developed publicly available sequences, acquisition protocols, and processing pipelines for whole-brain layer-fMRI. These protocols are validated across 60 hours of scanning in nine participants. Specifically, we identified and exploited methodological advancements for maximizing tSNR efficiency and test-retest reliability. We are sharing an extensive multi-modal whole-brain layer-fMRI dataset (20 scan hours of movie-watching in a single participant) for the purpose of benchmarking future method developments: The Kenshu dataset. With this dataset, we are also exemplifying the usefulness of whole brain layer-fMRI for commonly applied analysis approaches in modern cognitive neuroscience fMRI studies. This includes connectivity analyses, representational similarity matrix estimations, general linear model analyses, principal component analysis clustering, etc. We believe that this work paves the road for future routine measurements of directional functional connectivity across the entire brain.

Layers and columns are the dominant processing units in the human (neo)cortex at the mesoscopic scale. While the blood oxygenation dependent (BOLD) signal has a high detection sensitivity, it is biased towards unwanted signals from large draining veins at the cortical surface. The additional fMRI contrast of vascular space occupancy (VASO) has the potential to augment the neuroscientific interpretability of layer-fMRI results by means of capturing complementary information of locally specific changes in cerebral blood volume (CBV). Specifically, VASO is not subject to unwanted sensitivity amplifications of large draining veins. Because of constrained sampling efficiency, it has been mainly applied in combination with efficient block task designs and long trial durations. However, to study cognitive processes in neuroscientific contexts, or probe vascular reactivity, short stimulation periods are often necessary. Here, we developed a VASO acquisition procedure with a short acquisition period (895 ms volume acquisition) and sub-millimetre resolution. During visual event-related stimulation, we show reliable responses in visual cortices within a reasonable number of trials (∼20). Furthermore, the short TR and high spatial specificity of our VASO implementation enabled us to show differences in laminar reactivity and onset times. Finally, we explore the generalizability to a different stimulus modality (somatosensation). With this, we showed that CBV-sensitive VASO provides the means to capture layer-specific haemodynamic responses with high spatio-temporal resolution and is able to be used with event-related paradigms.

Cerebral blood volume (CBV) has been shown to be a robust and important physiological parameter for quantitative interpretation of functional (f)MRI, capable of delivering highly localized mapping of neural activity. Indeed, with recent advances in ultra-high-field (>=7T) MRI hardware and associated sequence libraries, it has become possible to capture non-invasive CBV weighted fMRI signals across cortical layers. One of the most widely used approaches to achieve this (in humans) is through vascular-space-occupancy (VASO) fMRI. Unfortunately, the exact contrast mechanisms of layer-dependent VASO fMRI have not been validated and thus interpretation of such data is confounded. Here we cross-validate layer-dependent VASO fMRI contrast in a preclinical rat model using well established (but invasive) imaging methods in response to neuronal activation (somatosensory cortex) and respiratory challenge (hypercapnia). In particular VASO derived CBV measures are directly compared to concurrent measures of total haemoglobin changes from high resolution intrinsic optical imaging spectroscopy (OIS). Through direct comparison of response magnitude, across time, negligible changes in hematocrit ratio during activation (neuronal or vascular) are inferred. Quantified cortical layer profiling is demonstrated and in agreement between both VASO and contrast enhanced fMRI (using monocrystalline iron oxide nanoparticles, MION). Responses show high spatial localisation to layers of cortical excitatory and inhibitory processing independent of confounding large draining veins which hamper BOLD fMRI studies. While we find increased VASO based CBV reactivity (3.1 ± 1.2 fold increase) in humans compared to rats it is demonstrated that this reflects differences in stimulus design rather than confounds of the VASO signal source. Together, our findings confirm that the VASO contrast is indeed a reliable estimate of layer-specific CBV changes. This validation study increases the neuronal interpretability of human layer-dependent fMRI results and should supersede BOLD fMRI as the method of choice in neuroscience application studies.HighlightsOur goal is to validate layer-specific VASO fMRI with gold standard methodsLayer-specific VASO sequences are implemented for 7T imaging in humans and ratsComparisons of VASO, optical imaging, and MION confirm the expected contrast originSomatosensory stimulation in humans and rats reveal the same layer-fMRI signaturesWe confirm that VASO is a valid measure to estimate layer-specific neural activityGraphical abstract