CNS*2016 Workshop

Connectome: Structure and Large-Scale Dynamics


Studies of the connectome are re-shaping the field of neuroscience. Networks have become a ubiquitous language. This is certainly reflected in computational neuroscience, where more and more groups are addressing problems at the large scale. However, the number of open questions is growing rapidly, so it is timely for computational neuroscientists to both direct our attention to the most important issues, and to grow capacity to take advantage of the opportunities that are unfolding. The workshop will present and highlight some of the important recent contributions on the structure of the connectome and the large-scale dynamics that it supports. We expect to have two round table sessions (closing the morning and the afternoon sessions) in which discussion will take place with the specific aim of exposing and highlighting the main issues and the interfaces where quantitative skills (abundant among computational neuroscientists) can be successfully applied to address exceptional emerging problems.


When: 
This will be a one day workshop taking place on the 6th of July 2016.

Where:
Jeju Island, South Korea.

Workshop forms part of the 25th Annual Computational Neuroscience Meeting CNS 2016.


Program

9:00-9:10  Welcome

9:10-9:40  Changsong Zhou
Cost-efficiency trade-off in brain connectome

9:40-10:10  James A. Roberts
Role of geometry in determining brain structure and metastable wave dynamics

10:10-10:40  Coffee Break

10:40-11:10  Leonardo L. Gollo
Hubs, wiring cost, and the evolutionary landscape of the human connectome

11:10-11:40  Ben D. Fulcher
Gene expression and neural activity in the connectome

11:40-12:00  Morning round-table discussion

12:00-13:30  Lunch Break

13:30-14:00  Bratislav Mišić
Cooperative and competitive spreading dynamics on the human connectome

14:00-14:30  Selen Atasoy
Connectome harmonics: self-organizing principle behind brain’s functional networks

14:30-15:00  Andreas Spiegler
Selective activation of resting state networks following focal stimulation in a large-scale brain network model

15:00-15:20  Coffee Break

15:20-15:50  Paula Sanz-Leon
Integrating connectomes and neural fields: a realistic approach to large-scale brain modelling

15:50-16:20  Jorge F. Mejias
Large-scale communication: signal propagation and frequency-dependent interactions in cortical networks

16:20-16:45  Afternoon round-table discussion 



Abstracts

Selen Atasoy (UPF)
Connectome harmonics: self-organizing principle behind brain’s functional networks

A fundamental characteristic of human brain activity is spontaneous coherent oscillations among spatially distributed cortical areas, forming the resting state networks. These networks are thought to emerge from local cortical dynamics and cortico-cortical interactions constrained by the anatomical structure of the human brain—the human connectome. Although various computational models have explored the spontaneous emergence of such oscillatory networks, our understanding lacks a unified fundamental principle revealing a direct macroscopic description of the collective cortical dynamics.


Ben D. Fulcher (Monash)
Gene expression and neural activity in the connectome

Recently, large-scale spatial databases of gene transcription in the brain have been made available. Combined with the high-quality measurement of brain connectivity in humans and other animals, and the ability to simulate brain dynamics mediated by these connectomes, these data provide a unique opportunity to investigate whether the organization of the connectome has any measurable molecular-level signatures. I will discuss methods for combining connectomic and transcriptional data, and outline a range of applications to real datasets.


Leonardo L. Gollo (QIMRB)
Hubs, wiring cost, and the evolutionary landscape of the human connectome

The organization of the human connectome is informed by a dynamic trade-off between low wiring length and high topological complexity: That is, the brain resides near a global minimum in an evolutionary landscape that pits the wiring costs of its axonal projections against the computational advantages conferred by efficiency and complexity. However, the factors shaping this landscape, the nature of appropriate benchmarks and the impact of subtle random variants of brain networks remain open questions. We address these issues by studying wiring length, core elements of brain structure and graph-theory measures in reference graphs that preserve the basic spatial embedding of the brain while destroying additional topological properties. We first show that the presence of spatial hubs introduces a local, but not global wiring cost. Positioning hubs broadly through distributed cortical regions – and not simply close to the geometric centre of the brain does, however, confer an additional wiring cost which the human cortex globally minimizes. Although random perturbations of brain networks can reduce the wiring length of the brain’s rich club inter-hub connections, these perturbations also quickly disconnect inter-hemispheric links to prefrontal hubs and yield daughter networks that substantially differ from one another. As the variation of structure accumulates, strong peripheral connections progressively connect to central nodes, and hubs are shift toward the middle of the brain. Progressive randomization of spatially embedded brain networks leads to a topologically unstable intermediate regime. Together with effects on wiring length, we suggest that prefrontal hub disconnections and topological instabilities act as evolutionary influences on complex brain networks.


Jorge F. Mejias (NYU)
Large-scale communication: signal propagation and frequency-dependent interactions in cortical networks

Understanding communication between large-scale brain structures is currently a major challenge in neuroscience. In this talk, I will cover two of our most recent modeling efforts to approach this problem, grounding both models on a recently obtained anatomical data set of the macaque connectivity. In the first part, I will present a large-scale laminar model used to study the enhanced rhythmic activity that has been recently associated with interactions between cortical areas. Our model displays four incremental levels of description (local, inter-laminar, inter-areal and large-scale circuits), and explains multiple electrophysiological observations across levels, including the enhancement of gamma/alpha rhythms associated with feedforward/feedback cortical interactions and the emergence of dynamical functional hierarchies. In the second part, I will present a similar large-scale model used to investigate signal propagation across cortical networks. The stable propagation of asynchronous neural activity represents a long-standing challenge, since as a signal propagates across areas, it may either die out or explode. The analysis of our anatomically-constrained large-scale model reveals the existence of a dynamical regime in which stable signal propagation is highly improved (around 100-fold with respect to previous models). The underlying dynamical principle can be seen as an extension of the well-known "balanced amplification" mechanism from local circuits to large-scale networks.


Bratislav Mišić (Indiana)
Cooperative and competitive spreading dynamics on the human connectome

Increasingly detailed data on the network topology of neural circuits create a need for theoretical principles that explain how these networks shape neural communication. Here we use a model of cascade spreading to reveal architectural features of human brain networks that facilitate spreading. Using an anatomical brain network derived from high-resolution diffusion spectrum imaging (DSI), we investigate scenarios where perturbations initiated at seed nodes result in global cascades that interact either cooperatively or competitively. We find that hub regions and a backbone of pathways facilitate early spreading, while the shortest path structure of the connectome enables cooperative effects, accelerating the spread of cascades. Finally, competing cascades become integrated by converging on polysensory associative areas. These findings show that the organizational principles of brain networks shape global communication and facilitate integrative function.


James A. Roberts (QIMRB)
Role of geometry in determining brain structure and metastable wave dynamics

Several recent studies have explored the role of how the brain's spatial embedding affects its network properties. Put another way, the simple fact that the brain is a physical object endows it with network properties that do not require any additional special wiring rules. In the first part of this talk I will present a novel method for generating random surrogate networks that preserve the relationship between edge weights and distances. This is useful for assessing the extent to which measured network properties can be explained by simple random wiring under the constraint of being a spatial network. The second part of this talk explores the brain dynamics that emerge from a model of neural masses coupled according to the structural connectome. Rich metastable spatiotemporal patterns exist, which spontaneously transition from one pattern to the next. These transitions provide useful statistics to compare the model to data, and comparison with geometry-preserving random networks reveals a role for geometry in determining the brain's dynamical flexibility.


Paula Sanz-Leon (Sydney)
Integrating connectomes and neural fields: a realistic approach to large-scale brain modelling

Existing computational models of large-scale brain activity generally fail to include one or more of the following features: realistic brain geometry; spatial heterogeneity; finite propagation speed; or biophysically based parameterizations. In this talk I will present preliminary results on a unifying framework that integrates the properties of current, seemingly disparate, approaches to large-scale biophysical modelling. More specifically, I will show that combining neural fields and connectomes enables modelling and simulations of tissue heterogeneity and wave propagation on a folded cortex. This extended computational framework, developed upon The Virtual Brain, will allow for a more detailed understanding of spatiotemporal dynamics of brain activity and represents one of the most realistic approximations to the physics of a human brain.


Andreas Spiegler (AMU)
Selective activation of resting state networks following focal stimulation in a large-scale brain network model

When the brain is stimulated, for example, by sensory inputs or goal-oriented tasks, the brain initially responds with activities in specific areas. The subsequent pattern formation of functional networks is constrained by the structural connectivity of the brain. The extent to which information is processed over short- or long-range SC is unclear. Whole-brain models based on long-range connections, for example, can partly describe measured functional connectivity dynamics at rest. The effects of structural connectivity on the network response to stimulation were studied in a whole-brain network model comprising long- and short-range connections. The results suggest that the stimulus-induced brain activity, which may indicate information and cognitive processing, follows specific routes imposed by structural networks explaining the emergence of functional networks. For example, the stimulation of specific areas results in the activation of one or more resting state networks.


Changsong Zhou (HKBU)
Cost-efficiency trade-off in brain connectome

The primate connectome, which displays diverse regional connectivity profiles, is well organized to support both segregated and integrated brain function. However, the organization mechanism shaping the characteristic connectivity profiles and their relationship with functional requirements remain unclear. The primate brain is an organ with high energy consumption, and it is speculated that its wiring diagrams is shaped by energy economy and functional constraints. Here, we explored the influence of two competing constraints and additional advanced functional requirements using the optimal trade-off model between the neural wiring cost and the representative functional requirement for processing efficiency. We found that the primate connectome displays a cost-efficiency trade-off, and up to 67% of the connections can be recovered by the optimal combination of the two basic constraints, which also shapes the connectivity of most regions. The majority of the remaining 33% of connections violating the constraints are long-distance links, which are concentrated on few cortical areas, termed long-distance connectors (LDCs). The LDCs are mainly non-hubs, but form a densely connected group overlapping on spatially segregated functional modalities. LDCs are essential components of the hierarchical modular architecture of the cortex, which are crucial for functional segregation and integration across different scales. These  organization features revealed by the optimization analysis provide evidence that the cortical connectome may violate a simple cost-efficiency trade-off to meet the demands of advanced functional segregation and integration among spatially distributed regions. These features also shed light on inherent vulnerabilities of brain networks in diseases and disorders.




Organizers

Leonardo L. Gollo, James A. Roberts  

(QIMR Berghofer Medical Research Institute, Australia).








Comments