Similar somatotopy for active and passive digit representation in primary somatosensory cortex.

Scientists traditionally use passive stimulation to examine the organisation of primary somatosensory cortex (SI). However, given the close, bidirectional relationship between the somatosensory and motor systems, active paradigms involving free movement may uncover alternative SI representational motifs. Here, we used 7 Tesla functional magnetic resonance imaging to compare hallmark features of SI digit representation between active and passive tasks which were unmatched on task or stimulus properties. The spatial location of digit maps, somatotopic organisation, and inter-digit representational structure were largely consistent between tasks, indicating representational consistency. We also observed some task differences. The active task produced higher univariate activity and multivariate representational information content (inter-digit distances). The passive task showed a trend towards greater selectivity for digits versus their neighbours. Our findings highlight that, while the gross features of SI functional organisation are task invariant, it is important to also consider motor contributions to digit representation.

Highly accurate retinotopic maps of the physiological blind spot in human visual cortex.

The physiological blind spot is a naturally occurring scotoma corresponding with the optic disc in the retina of each eye. Even during monocular viewing, observers are usually oblivious to the scotoma, in part because the visual system extrapolates information from the surrounding area. Unfortunately, studying this visual field region with neuroimaging has proven difficult, as it occupies only a small part of retinotopic cortex. Here, we used functional magnetic resonance imaging and a novel data-driven method for mapping the retinotopic organization in and around the blind spot representation in V1. Our approach allowed for highly accurate reconstructions of the extent of an observer's blind spot, and out-performed conventional model-based analyses. This method opens exciting opportunities to study the plasticity of receptive fields after visual field loss, and our data add to evidence suggesting that the neural circuitry responsible for impressions of perceptual completion across the physiological blind spot most likely involves regions of extrastriate cortex-beyond V1.

Predicting the retinotopic organization of human visual cortex from anatomy using geometric deep learning.

Whether it be in a single neuron or a more complex biological system like the human brain, form and function are often directly related. The functional organization of human visual cortex, for instance, is tightly coupled with the underlying anatomy with cortical shape having been shown to be a useful predictor of the retinotopic organization in early visual cortex. Although the current state-of-the-art in predicting retinotopic maps is able to account for gross individual differences, such models are unable to account for any idiosyncratic differences in the structure-function relationship from anatomical information alone due to their initial assumption of a template. Here we developed a geometric deep learning model capable of exploiting the actual structure of the cortex to learn the complex relationship between brain function and anatomy in human visual cortex such that more realistic and idiosyncratic maps could be predicted. We show that our neural network was not only able to predict the functional organization throughout the visual cortical hierarchy, but that it was also able to predict nuanced variations across individuals. Although we demonstrate its utility for modeling the relationship between structure and function in human visual cortex, our approach is flexible and well-suited for a range of other applications involving data structured in non-Euclidean spaces.

Manipulating the structure of natural scenes using wavelets to study the functional architecture of perceptual hierarchies in the brain.

Functional neuroimaging experiments that employ naturalistic stimuli (natural scenes, films, spoken narratives) provide insights into cognitive function "in the wild". Natural stimuli typically possess crowded, spectrally dense, dynamic, and multimodal properties within a rich multiscale structure. However, when using natural stimuli, various challenges exist for creating parametric manipulations with tight experimental control. Here, we revisit the typical spectral composition and statistical dependences of natural scenes, which distinguish them from abstract stimuli. We then demonstrate how to selectively degrade subtle statistical dependences within specific spatial scales using the wavelet transform. Such manipulations leave basic features of the stimuli, such as luminance and contrast, intact. Using functional neuroimaging of human participants viewing degraded natural images, we demonstrate that cortical responses at different levels of the visual hierarchy are differentially sensitive to subtle statistical dependences in natural images. This demonstration supports the notion that perceptual systems in the brain are optimally tuned to the complex statistical properties of the natural world. The code to undertake these stimulus manipulations, and their natural extension to dynamic natural scenes (films), is freely available.

Bayesian population receptive field modeling in human somatosensory cortex.

Somatosensation is fundamental to our ability to sense our body and interact with the world. Our body is continuously sampling the environment using a variety of receptors tuned to different features, and this information is routed up to primary somatosensory cortex. Strikingly, the spatial organization of the peripheral receptors in the body are well maintained, with the resulting representation of the body in the brain being referred to as the somatosensory homunculus. Recent years have seen considerable advancements in the field of high-resolution fMRI, which have enabled an increasingly detailed examination of the organization and properties of this homunculus. Here we combined advanced imaging techniques at ultra-high field (7T) with a recently developed Bayesian population receptive field (pRF) modeling framework to examine pRF properties in primary somatosensory cortex. In each subject, vibrotactile stimulation of the fingertips (i.e., the peripheral mechanoreceptors) modulated the fMRI response along the post-central gyrus and these signals were used to estimate pRFs. We found the pRF center location estimates to be in accord with previous work as well as evidence of other properties in line with the underlying neurobiology. Specifically, as expected from the known properties of cortical magnification, we find a larger representation of the index finger compared to the other stimulated digits (middle, index, little). We also show evidence that the little finger is marked by the largest pRF sizes, and that pRF size increases from anterior to posterior regions of S1. The ability to estimate somatosensory pRFs in humans provides an unprecedented opportunity to examine the neural mechanisms underlying somatosensation and is critical for studying how the brain, body, and environment interact to inform perception and action.

Using multi-echo simultaneous multi-slice (SMS) EPI to improve functional MRI of the subcortical nuclei of the basal ganglia at ultra-high field (7T).

The nuclei of the basal ganglia pose a special problem for functional MRI, especially at ultra-high field, because T2* variations between different regions result in suboptimal BOLD sensitivity when using gradient-echo echo-planar imaging (EPI). Specifically, the iron-rich lentiform nucleus of the basal ganglia, including the putamen and globus pallidus, suffers from substantial signal loss when imaging is performed using conventional single-echo EPI with echo times optimized for the cortex. Multi-echo EPI acquires several echoes at different echo times for every imaging slice, allowing images to be reconstructed with a weighting of echo times that is optimized individually for each voxel according to the underlying tissue or T2* properties. Here we show that multi-echo simultaneous multi-slice (SMS) EPI can improve functional activation of iron-rich subcortical regions while maintaining sensitivity within cortical areas. Functional imaging during a motor task known to elicit strong activations in the cortex and the subcortex (basal ganglia) was performed to compare the performance of multi-echo SMS EPI to single-echo SMS EPI. Notably within both the caudate nucleus and putamen of the basal ganglia, multi-echo SMS EPI yielded higher tSNR (an average 84% increase) and CNR (an average 58% increase), an approximate 3-fold increase in supra-threshold voxels, and higher t-values (an average 39% increase). The degree of improvement in the group level t-statistics was negatively correlated to the underlying T2* of the voxels, such that the shorter the T2*, as in the iron-rich nuclei of the basal ganglia, the higher the improvement of t-values in the activated region.

Serial correlations in single-subject fMRI with sub-second TR.

When performing statistical analysis of single-subject fMRI data, serial correlations need to be taken into account to allow for valid inference. Otherwise, the variability in the parameter estimates might be under-estimated resulting in increased false-positive rates. Serial correlations in fMRI data are commonly characterized in terms of a first-order autoregressive (AR) process and then removed via pre-whitening. The required noise model for the pre-whitening depends on a number of parameters, particularly the repetition time (TR). Here we investigate how the sub-second temporal resolution provided by simultaneous multislice (SMS) imaging changes the noise structure in fMRI time series. We fit a higher-order AR model and then estimate the optimal AR model order for a sequence with a TR of less than 600 ms providing whole brain coverage. We show that physiological noise modelling successfully reduces the required AR model order, but remaining serial correlations necessitate an advanced noise model. We conclude that commonly used noise models, such as the AR(1) model, are inadequate for modelling serial correlations in fMRI using sub-second TRs. Rather, physiological noise modelling in combination with advanced pre-whitening schemes enable valid inference in single-subject analysis using fast fMRI sequences.

Measuring the effects of attention to individual fingertips in somatosensory cortex using ultra-high field (7T) fMRI.

Attention to sensory information has been shown to modulate the neuronal processing of that information. For example, visuospatial attention acts by modulating responses at retinotopically appropriate regions of visual cortex (Puckett and DeYoe, 2015; Tootell et al. 1998). Much less, however, is known about the neuronal processing associated with attending to other modalities of sensory information. One reason for this is that visual cortex is relatively large, and therefore easier to access non-invasively in humans using tools such as functional magnetic resonance imaging (fMRI). With high-resolution fMRI, however, it is now possible to access smaller cortical areas such as primary somatosensory cortex (Martuzzi et al., 2014; Sanchez-Panchuelo et al., 2010; Schweisfurth et al. 2014; Schweizer et al. 2008). Here, we combined a novel experimental design and high-resolution fMRI at ultra-high field (7T) to measure the effects of attention to tactile stimulation in primary somatosensory cortex, S1. We find that attention modulates somatotopically appropriate regions of S1, and importantly, that this modulation can be measured at the level of the cortical representation of individual fingertips.

The spatiotemporal hemodynamic response function for depth-dependent functional imaging of human cortex.

The gray matter of human cortex is characterized by depth-dependent differences in neuronal activity and connections (Shipp, 2007) as well as in the associated vasculature (Duvernoy et al., 1981). The resolution limit of functional magnetic resonance imaging (fMRI) measurements is now below a millimeter, promising the non-invasive measurement of these properties in awake and behaving humans (Muckli et al., 2015; Olman et al., 2012; Ress et al., 2007). To advance this endeavor, we present a detailed spatiotemporal hemodynamic response function (HRF) reconstructed through the use of high-resolution, submillimeter fMRI. We decomposed the HRF into directions tangential and perpendicular to the cortical surface and found that key spatial properties of the HRF change significantly with depth from the cortical surface. Notably, we found that the spatial spread of the HRF increases linearly from 4.8mm at the gray/white matter boundary to 6.6mm near the cortical surface. Using a hemodynamic model, we posit that this effect can be explained by the depth profile of the cortical vasculature, and as such, must be taken into account to properly estimate the underlying neuronal responses at different cortical depths.

The attentional field revealed by single-voxel modeling of fMRI time courses.

The spatial topography of visual attention is a distinguishing and critical feature of many theoretical models of visuospatial attention. Previous fMRI-based measurements of the topography of attention have typically been too crude to adequately test the predictions of different competing models. This study demonstrates a new technique to make detailed measurements of the topography of visuospatial attention from single-voxel, fMRI time courses. Briefly, this technique involves first estimating a voxel's population receptive field (pRF) and then "drifting" attention through the pRF such that the modulation of the voxel's fMRI time course reflects the spatial topography of attention. The topography of the attentional field (AF) is then estimated using a time-course modeling procedure. Notably, we are able to make these measurements in many visual areas including smaller, higher order areas, thus enabling a more comprehensive comparison of attentional mechanisms throughout the full hierarchy of human visual cortex. Using this technique, we show that the AF scales with eccentricity and varies across visual areas. We also show that voxels in multiple visual areas exhibit suppressive attentional effects that are well modeled by an AF having an enhancing Gaussian center with a suppressive surround. These findings provide extensive, quantitative neurophysiological data for use in modeling the psychological effects of visuospatial attention.

An investigation of positive and inverted hemodynamic response functions across multiple visual areas.

Recent studies have demonstrated significant regional variability in the hemodynamic response function (HRF), highlighting the difficulty of correctly interpreting functional MRI (fMRI) data without proper modeling of the HRF. The focus of this study was to investigate the HRF variability within visual cortex. The HRF was estimated for a number of cortical visual areas by deconvolution of fMRI blood oxygenation level dependent (BOLD) responses to brief, large-field visual stimulation. Significant HRF variation was found across visual areas V1, V2, V3, V4, VO-1,2, V3AB, IPS-0,1,2,3, LO-1,2, and TO-1,2. Additionally, a subpopulation of voxels was identified that exhibited an impulse response waveform that was similar, but not identical, to an inverted version of the commonly described and modeled positive HRF. These voxels were found within the retinotopic confines of the stimulus and were intermixed with those showing positive responses. The spatial distribution and variability of these HRFs suggest a vascular origin for the inverted waveforms. We suggest that the polarity of the HRF is a separate factor that is independent of the suppressive or activating nature of the underlying neuronal activity. Correctly modeling the polarity of the HRF allows one to recover an estimate of the underlying neuronal activity rather than discard the responses from these voxels on the assumption that they are artifactual. We demonstrate this approach on phase-encoded retinotopic mapping data as an example of the benefits of accurately modeling the HRF during the analysis of fMRI data.