The BigBrainWarp toolbox for integration of BigBrain 3D histology with multimodal neuroimaging

Neuroimaging stands to benefit from emerging ultrahigh-resolution 3D histological atlases of the human brain; the first of which is 'BigBrain'. Here, we review recent methodological advances for the integration of BigBrain with multi-modal neuroimaging and introduce a toolbox, 'BigBrainWarp', that combines these developments. The aim of BigBrainWarp is to simplify workflows and support the adoption of best practices. This is accomplished with a simple wrapper function that allows users to easily map data between BigBrain and standard MRI spaces. The function automatically pulls specialised transformation procedures, based on ongoing research from a wide collaborative network of researchers. Additionally, the toolbox improves accessibility of histological information through dissemination of ready-to-use cytoarchitectural features. Finally, we demonstrate the utility of BigBrainWarp with three tutorials and discuss the potential of the toolbox to support multi-scale investigations of brain organisation.

A Simulation Toolkit for Testing the Sensitivity and Accuracy of Corticometry Pipelines

In recent years, the replicability of neuroimaging findings has become an important concern to the research community. Neuroimaging pipelines consist of myriad numerical procedures, which can have a cumulative effect on the accuracy of findings. To address this problem, we propose a method for simulating artificial lesions in the brain in order to estimate the sensitivity and specificity of lesion detection, using different automated corticometry pipelines. We have applied this method to different versions of two widely used neuroimaging pipelines (CIVET and FreeSurfer), in terms of coefficients of variation; sensitivity and specificity of detecting lesions in 4 different regions of interest in the cortex, while introducing variations to the lesion size, the blurring kernel used prior to statistical analyses, and different thickness metrics (in CIVET). These variations are tested in a between-subject design (in two random groups, with and without lesions, using T1-weigted MRIs of 152 individuals from the International Consortium of Brain Mapping (ICBM) dataset) and in a within-subject pre-/post-lesion design [using 21 T1-Weighted MRIs of a single adult individual, scanned in the Infant Brain Imaging Study (IBIS)]. The simulation method is sensitive to partial volume effect and lesion size. Comparisons between pipelines illustrate the ability of this method to uncover differences in sensitivity and specificity of lesion detection. We propose that this method be adopted in the workflow of software development and release.

Influence of Processing Pipeline on Cortical Thickness Measurement

In recent years, replicability of neuroscientific findings, specifically those concerning correlates of morphological properties of gray matter (GM), have been subject of major scrutiny. Use of different processing pipelines and differences in their estimates of the macroscale GM may play an important role in this context. To address this issue, here, we investigated the cortical thickness estimates of three widely used pipelines. Based on analyses in two independent large-scale cohorts, we report high levels of within-pipeline reliability of the absolute cortical thickness-estimates and comparable spatial patterns of cortical thickness-estimates across all pipelines. Within each individual, absolute regional thickness differed between pipelines, indicating that in-vivo thickness measurements are only a proxy of actual thickness of the cortex, which shall only be compared within the same software package and thickness estimation technique. However, at group level, cortical thickness-estimates correlated strongly between pipelines, in most brain regions. The smallest between-pipeline correlations were observed in para-limbic areas and insula. These regions also demonstrated the highest interindividual variability and the lowest reliability of cortical thickness-estimates within each pipeline, suggesting that structural variations within these regions should be interpreted with caution.

The genetic architecture of the human cerebral cortex

The cerebral cortex underlies our complex cognitive capabilities, yet little is known about the specific genetic loci that influence human cortical structure. To identify genetic variants that affect cortical structure, we conducted a genome-wide association meta-analysis of brain magnetic resonance imaging data from 51,665 individuals. We analyzed the surface area and average thickness of the whole cortex and 34 regions with known functional specializations. We identified 199 significant loci and found significant enrichment for loci influencing total surface area within regulatory elements that are active during prenatal cortical development, supporting the radial unit hypothesis. Loci that affect regional surface area cluster near genes in Wnt signaling pathways, which influence progenitor expansion and areal identity. Variation in cortical structure is genetically correlated with cognitive function, Parkinson's disease, insomnia, depression, neuroticism, and attention deficit hyperactivity disorder.

Cortical surface area variations within the dorsolateral prefrontal cortex are better predictors of future cognitive performance than fluid ability and working memory

BACKGROUND

Are cognitive and biological variables useful for predicting future behavioral outcomes?

METHOD

In two independent groups, we measured a set of cognitive (fluid and crystallized intelligence, working memory, and attention control) and biological (cortical thickness and cortical surface area) variables on two occasions separated by six months, to predict behavioral outcomes of interest (performance on an adaptive version of the n-back task) measured twelve and eighteen months later. We followed three stages: discovery, validation, and generalization. In the discovery stage, cognitive/biological variables and the behavioral outcome of interest were assessed in a group of individuals (in-sample). In the validation stage, the cognitive and biological variables were related with a parallel version of the behavioral outcome assessed several months later. In the generalization stage, the validation findings were tested in an independent group of individuals (out-of-sample).

RESULTS

The key finding revealed that cortical surface area variations within the right dorsolateral prefrontal cortex predict the behavioral outcome of interest in both groups, whereas the cognitive variables failed to show reliable predictive validity.

CONCLUSIONS

Individual differences in biological variables might predict future behavioral outcomes better than cognitive variables concurrently correlated with these behavioral outcomes.

Partial correlation analysis reveals abnormal retinotopically organized functional connectivity of visual areas in amblyopia

Amblyopia is a prevalent developmental visual disorder of childhood that typically persists in adults. Due to altered visual experience during critical periods of youth, the structure and function of adult visual cortex is abnormal. In addition to substantial deficits shown with task-based fMRI, previous studies have used resting state measures to demonstrate altered long-range connectivity in amblyopia. This is the first study in amblyopia to analyze connectivity between regions of interest that are smaller than a single cortical area and to apply partial correlation analysis to reduce network effects. We specifically assess short-range connectivity between retinotopically defined regions of interest within the occipital lobe of 8 subjects with amblyopia and 7 subjects with normal vision (aged 19–45). The representations of visual areas V1, V2, and V3 within each of the four quadrants of visual space were further subdivided into three regions based on maps of visual field eccentricity. Connectivity between pairs of all nine regions of interest in each quadrant was tested via correlation and partial correlation for both groups. Only the tests of partial correlation, i.e., correlation between time courses of two regions following the regression of time courses from all other regions, yielded significant differences between resting state functional connectivity in amblyopic and normal subjects. Subjects with amblyopia showed significantly higher partial correlation between para-foveal and more eccentric representations within V1, and this effect associated with poor acuity of the worse eye. In addition, we observed reduced correlation in amblyopic subjects between isoeccentricity regions in V1 and V2, and separately, between such regions in V2 and V3. We conclude that partial correlation-based connectivity is altered in an eccentricity-dependent pattern in visual field maps of amblyopic patients. Moreover, results are consistent with known clinical and psychophysical vision loss. More broadly, this provides evidence that abnormal cortical adaptations to disease may be better isolated with tests of partial correlation connectivity than with the regular correlation techniques that are currently widely used.

•

Cortical functional connectivity abnormalities exist in amblyopia at a scale finer than previously reported.

•

Connectivity changes within primary visual cortex are consistent with known loss of function.

•

Connectivity changes between visual areas are consistent with concept of deafferentation.

•

Partial correlation differentiates patients from controls, whereas correlation does not.

Partial Correlation-Based Retinotopically Organized Resting-State Functional Connectivity Within and Between Areas of the Visual Cortex Reflects More Than Cortical Distance

Numerous studies have demonstrated functional magnetic resonance imaging (fMRI)-based resting-state functional connectivity (RSFC) between cortical areas. Recent evidence suggests that synchronous fluctuations in blood oxygenation level-dependent fMRI reflect functional organization at a scale finer than that of visual areas. In this study, we investigated whether RSFCs within and between lower visual areas are retinotopically organized and whether retinotopically organized RSFC merely reflects cortical distance. Subjects underwent retinotopic mapping and separately resting-state fMRI. Visual areas V1, V2, and V3, were subdivided into regions of interest (ROIs) according to quadrants and visual field eccentricity. Functional connectivity (FC) was computed based on Pearson's linear correlation (correlation), and Pearson's linear partial correlation (correlation between two time courses after the time courses from all other regions in the network are regressed out). Within a quadrant, within visual areas, all correlation and nearly all partial correlation FC measures showed statistical significance. Consistently in V1, V2, and to a lesser extent in V3, correlation decreased with increasing eccentricity separation. Consistent with previously reported monkey anatomical connectivity, correlation/partial correlation values between regions from adjacent areas (V1-V2 and V2-V3) were higher than those between nonadjacent areas (V1-V3). Within a quadrant, partial correlation showed consistent significance between regions from two different areas with the same or adjacent eccentricities. Pairs of ROIs with similar eccentricity showed higher correlation/partial correlation than pairs distant in eccentricity. Between dorsal and ventral quadrants, partial correlation between common and adjacent eccentricity regions within a visual area showed statistical significance; this extended to more distant eccentricity regions in V1. Within and between quadrants, correlation decreased approximately linearly with increasing distances separating the tested ROIs. Partial correlation showed a more complex dependence on cortical distance: it decreased exponentially with increasing distance within a quadrant, but was best fit by a quadratic function between quadrants. We conclude that RSFCs within and between lower visual areas are retinotopically organized. Correlation-based FC is nonselectively high across lower visual areas, even between regions that do not share direct anatomical connections. The mechanisms likely involve network effects caused by the dense anatomical connectivity within this network and projections from higher visual areas. FC based on partial correlation, which minimizes network effects, follows expectations based on direct anatomical connections in the monkey visual cortex better than correlation. Last, partial correlation-based retinotopically organized RSFC reflects more than cortical distance effects.

Gray matter responsiveness to adaptive working memory training: a surface-based morphometry study

Here we analyze gray matter indices before and after completing a challenging adaptive cognitive training program based on the n-back task. The considered gray matter indices were cortical thickness (CT) and cortical surface area (CSA). Twenty-eight young women (age range 17-22 years) completed 24 training sessions over the course of 3 months (12 weeks, 24 sessions), showing expected performance improvements. CT and CSA values for the training group were compared with those of a matched control group. Statistical analyses were computed using a ROI framework defined by brain areas distinguished by their genetic underpinning. The interaction between group and time was analyzed. Middle temporal, ventral frontal, inferior parietal cortices, and pars opercularis were the regions where the training group showed conservation of gray matter with respect to the control group. These regions support working memory, resistance to interference, and inhibition. Furthermore, an interaction with baseline intelligence differences showed that the expected decreasing trend at the biological level for individuals showing relatively low intelligence levels at baseline was attenuated by the completed training.

Reproducibility of neuroimaging analyses across operating systems

Neuroimaging pipelines are known to generate different results depending on the computing platform where they are compiled and executed. We quantify these differences for brain tissue classification, fMRI analysis, and cortical thickness (CT) extraction, using three of the main neuroimaging packages (FSL, Freesurfer and CIVET) and different versions of GNU/Linux. We also identify some causes of these differences using library and system call interception. We find that these packages use mathematical functions based on single-precision floating-point arithmetic whose implementations in operating systems continue to evolve. While these differences have little or no impact on simple analysis pipelines such as brain extraction and cortical tissue classification, their accumulation creates important differences in longer pipelines such as subcortical tissue classification, fMRI analysis, and cortical thickness extraction. With FSL, most Dice coefficients between subcortical classifications obtained on different operating systems remain above 0.9, but values as low as 0.59 are observed. Independent component analyses (ICA) of fMRI data differ between operating systems in one third of the tested subjects, due to differences in motion correction. With Freesurfer and CIVET, in some brain regions we find an effect of build or operating system on cortical thickness. A first step to correct these reproducibility issues would be to use more precise representations of floating-point numbers in the critical sections of the pipelines. The numerical stability of pipelines should also be reviewed.

BigBrain: an ultrahigh-resolution 3D human brain model

Reference brains are indispensable tools in human brain mapping, enabling integration of multimodal data into an anatomically realistic standard space. Available reference brains, however, are restricted to the macroscopic scale and do not provide information on the functionally important microscopic dimension. We created an ultrahigh-resolution three-dimensional (3D) model of a human brain at nearly cellular resolution of 20 micrometers, based on the reconstruction of 7404 histological sections. "BigBrain" is a free, publicly available tool that provides considerable neuroanatomical insight into the human brain, thereby allowing the extraction of microscopic data for modeling and simulation. BigBrain enables testing of hypotheses on optimal path lengths between interconnected cortical regions or on spatial organization of genetic patterning, redefining the traditional neuroanatomy maps such as those of Brodmann and von Economo.

Evaluation and calibration of functional network modeling methods based on known anatomical connections

Recent studies have identified large scale brain networks based on the spatio-temporal structure of spontaneous fluctuations in resting-state fMRI data. It is expected that functional connectivity based on resting-state data is reflective of - but not identical to - the underlying anatomical connectivity. However, which functional connectivity analysis methods reliably predict the network structure remains unclear. Here we tested and compared network connectivity analysis methods by applying them to fMRI resting-state time-series obtained from the human visual cortex. The methods evaluated here are those previously tested against simulated data in Smith et al. (Neuroimage, 2011). To this end, we defined regions within retinotopic visual areas V1, V2, and V3 according to their eccentricity in the visual field, delineating central, intermediate, and peripheral eccentricity regions of interest (ROIs). These ROIs served as nodes in the models we study. We based our evaluation on the "ground-truth", thoroughly studied retinotopically-organized anatomical connectivity in the monkey visual cortex. For each evaluated method, we computed the fractional rate of detecting connections known to exist ("c-sensitivity"), while using a threshold of the 95th percentile of the distribution of interaction magnitudes of those connections not expected to exist. Under optimal conditions - including session duration of 68min, a relatively small network consisting of 9 nodes and artifact-free regression of the global effect - each of the top methods predicted the expected connections with 67-85% c-sensitivity. Correlation methods, including Correlation (Corr; 85%), Regularized Inverse Covariance (ICOV; 84%) and Partial Correlation (PCorr; 81%) performed best, followed by Patel's Kappa (80%), Bayesian Network method PC (BayesNet; 77%), General Synchronization measures (67-77%), and Coherence (CohB; 74%). With decreased session duration, these top methods saw decreases in c-sensitivities, achieving 59-76% for 17min sessions. With a short resting-state fMRI scan of 8.5min, none of the methods predicted the real network well, with Corr (65%) performing best. With increased complexity of the network from 9 to 36 nodes, multivariate methods including PCorr and BayesNet saw a decrease in performance. Artifact-free regression of the global effect increased the c-sensitivity of the top-performing methods. In an overall evaluation across all tests we performed, correlation methods (Corr, ICOV, and PCorr), Patel's Kappa, and BayesNet method PC set themselves somewhat above all other methods. We propose that data-based calibration based on known anatomical connections be integrated into future network studies, in order to maximize sensitivity and reduce false positives.

Mechanisms of cross-modal plasticity in early-blind subjects

A variety of studies have demonstrated enhanced blood oxygenation level dependent responses to auditory and tactile stimuli within occipital cortex as a result of early blindness. However, little is known about the organizational principles that drive this cross-modal plasticity. We compared BOLD responses to a wide variety of auditory and tactile tasks (vs. rest) in early-blind and sighted subjects. As expected, cross-modal responses were larger in blind than in sighted subjects in occipital cortex for all tasks (cross-modal plasticity). Within both blind and sighted subject groups, we found patterns of cross-modal activity that were remarkably similar across tasks: a large proportion of cross-modal responses within occipital cortex are neither task nor stimulus specific. We next examined the mechanisms underlying enhanced BOLD responses within early-blind subjects. We found that the enhancement of cross-modal responses due to early blindness was best described as an additive shift, suggesting that cross-modal plasticity within blind subjects does not originate from either a scaling or unmasking of cross-modal responsivities found in sighted subjects.

Integration of one- and two-dimensional motion signals in infants: evidence from the barber-pole illusion

Several previous studies in adults have investigated how one- and two-dimensional moving features are integrated into a coherent global motion percept by studying the "barber-pole illusion"; when a one-dimensional moving grating is presented within a rectangular aperture, the two-dimensional line terminators at the edges of the aperture bias the perceived direction of motion toward the longer axis of the aperture. In the current study, we used barber-pole stimuli to investigate the development of motion mechanisms that integrate one- and two-dimensional motion signals. Using a directional eye movement technique, we measured responses to obliquely moving gratings presented within horizontally vs. vertically oriented apertures, in infants (ages 2-5 months) and adults. For all ages, we found that horizontal eye movements were significantly stronger when gratings were presented within horizontal than within vertical apertures, as predicted by the barber-pole illusion. Additionally, we devised a way to infer the "effective shift" in eye movement direction produced by the barber-pole illusion. Using a simple motion integration model, effective shift values were then used to calculate the relative weightings of one- and two-dimensional motion signals to direction coding. The results show that by 2 months of age, infants integrate one- and two-dimensional motion signals, and that the relative weighting of one- and two-dimensional signals remains roughly constant from 2 months of age into adulthood.

Young children's overgeneralizations with fixed transitivity verbs

The present study examined English-speaking children's tendency to make argument structure overgeneralization errors (e.g., I disappeared it). Children were exposed to several English verbs of fixed transitivity (exclusively intransitive or exclusively transitive) and then asked questions that encouraged them to overgeneralize usage of the verbs. Seventy-two children (24 in each of three age groups: 3, 4/5, and 8 years of age) experienced four actions performed by puppets. Each action had two verbs of similar meaning associated with it in the context of the experimental action: one more familiar to young children and one less familiar. Children at all ages were more likely to overgeneralize usage of verbs that were less familiar to them, supporting the hypothesis that children's usage of verbs in particular construction types becomes entrenched over time. As children solidly learn the transitivity status of particular verbs, they become more reluctant to use those verbs in other argument structure constructions.

Effects of prosodic stress and serial position on syllable omission in first words

The authors documented syllable omission in one child's multisyllabic vocabulary from 10 to 20 months of age to evaluate L. Gerken's (1991, 1994a, 1994b) proposal that children organize their productions according to a trochaic metrical (strong-weak) template and omit syllables from target utterances that do not conform to this pattern. The trochaic template hypothesis was not supported by these early productions. Results indicated that the likelihood of producing a target syllable was influenced primarily by the strength of the prosodic stress placed on it and secondarily by its serial order within a word. Over time, the child demonstrated an increasing ability to include syllables with weaker prosodic stress in multisyllabic productions. Omissions became much less common with the onset of 2-word speech.

A production limitation in syllable number: a longitudinal study of one child's early vocabulary

The present paper reports on the phonological form of one child's productive vocabulary from age 0;10 to 1;8 with primary focus on his production of multisyllabic targets. A large percentage of his multisyllabic vocabulary was produced as one syllable until the age of 1;6. This limitation was not due to a tendency to extract only single syllables from the speech stream, but rather due primarily to a limitation on production. While some portion of his one-syllable productions could be interpreted as the result of single syllable extraction, a sizeable portion affirmed that he extracted the target size correctly by his inclusion of first and final target phonemes in his productions (e.g. [po] for piano and [kiz] for candies). The resolution of this limitation coincides with his move toward two-word speech. We conclude that there is a developmental and perhaps maturational limitation in the capacity to carry out the processes underlying word and sentence production.