BigBrain 3D atlas of cortical layers: Cortical and laminar thickness gradients diverge in sensory and motor cortices

Pre-Brodmann pioneers of cortical cytoarchitectonics II: Carl Hammarberg, Alfred Walter Campbell and Grafton Elliot Smith

The present study and the preceding paper revisit landmark discoveries that paved the way to the definition of the renowned Brodmann areas in the human cerebral cortex, in an attempt to rectify certain undeserved historical neglects. A 'second period of discoveries', from 1893 to 1908, is marked by the work of Carl Hammarberg (1865-1893) in Uppsala, Alfred Walter Campbell (1868-1937) in Liverpool and Grafton Elliot Smith (1871-1937) in Cairo. Their classical findings are placed in a modern perspective.

Convergence of cortical types and functional motifs in the human mesiotemporal lobe

The mesiotemporal lobe (MTL) is implicated in many cognitive processes, is compromised in numerous brain disorders, and exhibits a gradual cytoarchitectural transition from six-layered parahippocampal isocortex to three-layered hippocampal allocortex. Leveraging an ultra-high-resolution histological reconstruction of a human brain, our study showed that the dominant axis of MTL cytoarchitectural differentiation follows the iso-to-allocortical transition and depth-specific variations in neuronal density. Projecting the histology-derived MTL model to in-vivo functional MRI, we furthermore determined how its cytoarchitecture underpins its intrinsic effective connectivity and association to large-scale networks. Here, the cytoarchitectural gradient was found to underpin intrinsic effective connectivity of the MTL, but patterns differed along the anterior-posterior axis. Moreover, while the iso-to-allocortical gradient parametrically represented the multiple-demand relative to task-negative networks, anterior-posterior gradients represented transmodal versus unimodal networks. Our findings establish that the combination of micro- and macrostructural features allow the MTL to represent dominant motifs of whole-brain functional organisation.

Higher and deeper: Bringing layer fMRI to association cortex

Recent advances in fMRI have enabled non-invasive measurements of brain function in awake, behaving humans at unprecedented spatial resolutions, allowing us to separate activity in distinct cortical layers. While most layer fMRI studies to date have focused on primary cortices, we argue that the next big steps forward in our understanding of cognition will come from expanding this technology into higher-order association cortex, to characterize depth-dependent activity during increasingly sophisticated mental processes. We outline phenomena and theories ripe for investigation with layer fMRI, including perception and imagery, selective attention, and predictive coding. We discuss practical and theoretical challenges to cognitive applications of layer fMRI, including localizing regions of interest in the face of substantial anatomical heterogeneity across individuals, designing appropriate task paradigms within the confines of acquisition parameters, and generating hypotheses for higher-order brain regions where the laminar circuitry is less well understood. We consider how applying layer fMRI in association cortex may help inform computational models of brain function as well as shed light on consciousness and mental illness, and issue a call to arms to our fellow methodologists and neuroscientists to bring layer fMRI to this next frontier.

Different scaling of linear models and deep learning in UKBiobank brain images versus machine-learning datasets

Recently, deep learning has unlocked unprecedented success in various domains, especially using images, text, and speech. However, deep learning is only beneficial if the data have nonlinear relationships and if they are exploitable at available sample sizes. We systematically profiled the performance of deep, kernel, and linear models as a function of sample size on UKBiobank brain images against established machine learning references. On MNIST and Zalando Fashion, prediction accuracy consistently improves when escalating from linear models to shallow-nonlinear models, and further improves with deep-nonlinear models. In contrast, using structural or functional brain scans, simple linear models perform on par with more complex, highly parameterized models in age/sex prediction across increasing sample sizes. In sum, linear models keep improving as the sample size approaches ~10,000 subjects. Yet, nonlinearities for predicting common phenotypes from typical brain scans remain largely inaccessible to the examined kernel and deep learning methods.

The frequency gradient of human resting-state brain oscillations follows cortical hierarchies

The human cortex is characterized by local morphological features such as cortical thickness, myelin content, and gene expression that change along the posterior-anterior axis. We investigated if some of these structural gradients are associated with a similar gradient in a prominent feature of brain activity - namely the frequency of oscillations. In resting-state MEG recordings from healthy participants (N = 187) using mixed effect models, we found that the dominant peak frequency in a brain area decreases significantly along the posterior-anterior axis following the global hierarchy from early sensory to higher order areas. This spatial gradient of peak frequency was significantly anticorrelated with that of cortical thickness, representing a proxy of the cortical hierarchical level. This result indicates that the dominant frequency changes systematically and globally along the spatial and hierarchical gradients and establishes a new structure-function relationship pertaining to brain oscillations as a core organization that may underlie hierarchical specialization in the brain.

Closing the gap from transcription to the structural connectome enhances the study of connections in the human brain

The brain is composed of a complex web of networks but we have yet to map the structural connections of the human brain in detail. Diffusion MR imaging is a high-throughput method that relies on the principle of diffusion to reconstruct tracts (ie, pathways) across the brain. Although diffusion MR tractography is an exciting method to explore the structural connectivity of the brain in development and across species, the tractography has at times led to questionable interpretations. There are at present few if any alternative methods to trace structural pathways in the human brain. Given these limitations and the potential of diffusion MR imaging to map the human connectome, it is imperative that we develop new approaches to validate neuroimaging techniques. I discuss our recent studies integrating neuroimaging with transcriptional and anatomical variation across humans and other species over the course of development and in adulthood. Developing a novel framework to harness the potential of diffusion MR tractography provides new and exciting opportunities to study the evolution of developmental mechanisms generating variation in connections and bridge the gap between model systems to humans.

Predicting Neural Response Latency of the Human Early Visual Cortex from MRI-Based Tissue Measurements of the Optic Radiation

Although the non-invasive measurement of visually evoked responses has been extensively studied, the structural basis of variabilities in latency in healthy humans is not well understood. We investigated how tissue properties of optic radiation could predict interindividual variability in the latency of the initial visually evoked component (C1), which may originate from the primary visual cortex (V1). We collected C1 peak latency data using magnetoencephalography (MEG) and checkerboard stimuli, and multiple structural magnetic resonance imaging (MRI) data from 20 healthy subjects. While we varied the contrast and position of the stimuli, the C1 measurement was most reliable when high-contrast stimuli were presented to the lower visual field (LVF). We then attempted to predict interindividual variability in C1 peak latency in this stimulus condition with a multiple regression model using MRI parameters along the optic radiation. We found that this model could predict >20% of variance in C1 latency, when the data were averaged across the hemispheres. The model using the corticospinal tract did not predict variability in C1 latency, suggesting that there is no evidence for generalization to a non-visual tract. In conclusion, our results suggest that the variability in neural latencies in the early visual cortex in healthy subjects can be partly explained by tissue properties along the optic radiation. We discuss the challenges of predicting neural latency using current structural neuroimaging methods and other factors that may explain interindividual variance in neural latency.