IHAVTVB
Interaction between Human Atlas Viewer and The Virtual Brain
Project description
Scientific question: Brain regions in the HBP human brain atlas are backed by experimental data, that can provide important features like cell density, properties of connectivity, receptor architecture, and similar. How can we use such different features of each brain region from the atlas to inform and validate the large brain network dynamics?
Background and state of the art:
We use diverse approaches, including nonlinear dynamics, complex systems theory, probability theory and statistical mechanics, to study how brain activity arises through the complex interaction of numerous neurons. Theoretical models play an important role in bridging the gap between physical descriptions of cortical systems and empirical studies of brain functioning. We model brain activity using TVB (The Virtual Brain) at various physical scales using different levels of abstraction.
Currently, the models of local dynamics of brain network models implemented in TVB are parametrized uniformly for the whole brain, i.e. models of all brain areas are represented with models with identical parameters regarding the architecture and properties of neuronal cells as well as local connectivity at the macro- and microscale. Consequently, they show the same dynamical repertoire under the weak-coupling assumption. This approximation was necessary to build whole-brain network models up to now, due to the lack of consistent data to inform the spatially non-uniform parametrization on a whole-brain scale. With the ongoing development of HBP's multilevel human atlas, providing increasing access to high quality area- specific experimental data, we can now go beyond this approximation and explore different ways of mapping model parameters specified according to the anatomical information for different brain regions.
HBP Platform usage and novelty:
The project will act as a pilot user of the atlas services in HBP's Neuroinforrnatics platform, to incorporate atlas data directly in the simulation configuration in the TVB simulator. It will hereby reveal immediate practical requirements on atlas services and data components for linking atlases with simulation, point the HBP developers to possibly missing functionalities, and help to optimize the design of application programming interfaces.
The novelty of the proposed project lies in (i) spatial anchoring of the whole-brain mean-field model, and (ii) the realistic brain network model with brain region specified parameters. This work will, for the first time, enable scientists to show the initial evidence that spatial non-uniformity plays a role in shaping the brain's large-scale organization and the subsequent brain imaging signals.
Model Parameters:
In the case that we want to model spatially heterogeneous parameter of a brain model:
First, we explore the anatomical information from our research interested brain regions, then we click a button such as "go to TVB", which will input our interested anatomical information into the TVB. Inside TVB, we will map the given anatomical information to model parameters. Then we have a button which will lead LIS back to the Human atlas viewer, where we are able to see the overlay of the model parameters and other interest such as in the layer densities in the BigBrain anatomical template.
Model configuration:
The connectivity information is important for our brain models. In the case that we want to verify the connectivity: We set up the local connectivity kernel in TVB platform, then click the button 'go to Atlas' Then we could load the connectivity kernel visualizer in the online atlas viewer and search the area of interest for known low-level datasets describing connectivity of the ROI.
Dynamic features:
Once we obtain our simulation results, besides using visualizers in TVB, we also could click the button "GO to Atlas" to overlay the results to the online atlas. Then we could visualize the dynamic features such as frequency feature in a given brain region. By doing this, we are able to inspect the spatiotemporal patterns together with other datasets available in the viewer. Our scientific interesting on the brain-region-specific large-scale brain network models will better simulate the variety of data features from each brain areas in the empirical datasets. This will improve the currently widely used (and criticized) approximation of using the identical mean-field model of local dynamics for all brain areas.
The implementation of the bi-directional bridge between TVB and the multi-level atlas of the Human Brain will benefit the numerous users of the TVB modeling platform (2000* registered scientists, 167004 downloads). Also, our scientific objective to build the large-scale brain networks with brain-region specified model parameters will bring qualitative improvement to better simulate and interpret the empirical functional data.
Collaboration with HBP
The project will work towards a direct connection of HBP's neuroinformatics platoform with a multi-scale simulation engine to inform large brain network models with brain region specific experimental parameters. The parameter mapping will be based on brain-region specified data features, such as neuronal cell density for different cortical layers. These anatomical data features will be extracted from the HBP multi-level atlas to the degree that the HBP platform allows. After the parameters are derived automatically in the TVB, a detailed interactive inspection of their spatial distribution overlaid with other data available in the Multi-Level Atlas will be necessary. A workflow linked to HBP's interactive atlas should allow the user to switch between the configuration of the model within TVB, and exploring the data in the atlas for interested brain regions.
The aim will be to easily and fully use resources from both Human brain atlas viewers and the TVB platform. Then, when we are exploring a brain region, we can choose the specified data features for the models. Once we choose these data features, then these data features can be accessed by the TVB platform where we can choose the brain models and manipulate the model parameters for the specified brain regions. Once we have the brain dynamics, the ideal situation is that we can go back to view the data's dynamic feature on the human brain atlas viewers.
The project will imply to review the data provided through the HBP neuroinformatics platform, support in anchoring atlas data and clarifying ontologies, and development or extension of Python client libraries for HBP's atlas services, to match the workflow of large-scale modeling.
The software architecture of TVB is highly modular, separating the simulation engine from the model configuration and graphical user interface (Woodman et al., 2014), and the integration with the Multi-Level Atlas will be feasible with manageable software engineering effort, as both graphical user interfaces are browser-based.
All the components of TVB are open source and available on GitHub, together with complete pipeline to construct a personalized whole-brain models from neuroimaging data (Proix et al., 2016).
Key facts
Time frame: 2019-2020
Origin: Voucher Programme
Funding: University of Newcastle
