Modelling Human Brain Development
Lead Research Organisation:
Newcastle University
Department Name: Computing Sciences
Abstract
The neural network of the human brain is arguably the most complex biological pattern; however, the mechanisms forming such neural systems are unclear. Neural systems show structurally emergent properties in terms of their topology and functionally emergent properties concerning information processing. Structural properties are the rise of modular and hierarchical connectivity, of long-distance connections, and of highly connected nodes. Functional properties are the distribution and integration of information, the formation of specialized modules dealing with different tasks, and a rapid reaction time due to parallel processing.
Recent advances in neuroimaging, using diffusion tensor imaging, allow us to observe how the human brain network differs over ages ranging from the embryonic to the adult stage (age of 20 years). This project will analyse how the human brain network arises during development by combining data analysis with simulations of brain development. Objectives are to develop a simulation of human brain development, to analyse network features of human brains at different developmental stages, and to compare simulations with real data to discover the underlying mechanisms for brain network development. Simulations are crucial to study the role of different developmental parameters on the final brain network as well as on intermediate networks during development. Understanding how parameters lead to (adult) network features will help to evaluate the contribution of these parameters to healthy and pathological development. Once understanding the time course of development, we should also be able to predict the probabilities of future stages of development. This will be crucial for giving a prognosis for the progression of developmental diseases. In addition to understanding the formation of human cognitive systems, these results will inform the design and update of artificial information processing systems.
Identifying these key developmental mechanisms will greatly improve our understanding of emergence in biological systems. In addition, it might lead to several predictions about the rise of brain disorders such as schizophrenia, epilepsy, and autism that often originate during development and are linked to changes in hub organization. Beyond biological pattern formation, such 'algorithms' for human brain development could inform us how to build artificial intelligent systems. Rather than constructing artificial brains in a top-down manner, applying identified mechanisms for neural network development will allow the emergence of intelligent information processing systems leading to systems that are more adaptable. In summary, the formation of human brains is a fundamental question that touches on the emergence of natural and man-made information processing systems.
Recent advances in neuroimaging, using diffusion tensor imaging, allow us to observe how the human brain network differs over ages ranging from the embryonic to the adult stage (age of 20 years). This project will analyse how the human brain network arises during development by combining data analysis with simulations of brain development. Objectives are to develop a simulation of human brain development, to analyse network features of human brains at different developmental stages, and to compare simulations with real data to discover the underlying mechanisms for brain network development. Simulations are crucial to study the role of different developmental parameters on the final brain network as well as on intermediate networks during development. Understanding how parameters lead to (adult) network features will help to evaluate the contribution of these parameters to healthy and pathological development. Once understanding the time course of development, we should also be able to predict the probabilities of future stages of development. This will be crucial for giving a prognosis for the progression of developmental diseases. In addition to understanding the formation of human cognitive systems, these results will inform the design and update of artificial information processing systems.
Identifying these key developmental mechanisms will greatly improve our understanding of emergence in biological systems. In addition, it might lead to several predictions about the rise of brain disorders such as schizophrenia, epilepsy, and autism that often originate during development and are linked to changes in hub organization. Beyond biological pattern formation, such 'algorithms' for human brain development could inform us how to build artificial intelligent systems. Rather than constructing artificial brains in a top-down manner, applying identified mechanisms for neural network development will allow the emergence of intelligent information processing systems leading to systems that are more adaptable. In summary, the formation of human brains is a fundamental question that touches on the emergence of natural and man-made information processing systems.
Planned Impact
This work could lead to insights into the normal and pathological development of the human brain. We will publish the results in experimental and clinical journals to ensure that they are visible to practitioners in the field. The articles will also be open-access so that researchers from both areas, computing and the life sciences, will be able to see the results. We will discuss clinical applications of this information with Dr Jonathan Richardson who is leader of the psychiatry health informatics group in the UK. In particular, we will evaluate whether we can release our software as a clinical tool for evaluating patients or at-risk subjects for developmental brain disorders. Subjects with high-risk include children and young adults with relatives who suffered from developmental brain disorders (e.g. schizophrenia). We are aware of the legal and administrative procedures for using software within the NHS and were already in contact with Clarity Informatics, a Newcastle company providing data analysis solutions for the UK healthcare system.
This project might also inform the design of technical communication and information processing systems. We will therefore discuss applications with our partners in the IT industry. In particular, we will use our existing links with Dr Kozloski (IBM T.J. Watson Research Center, New York) and Prof Steve Furber (developer of ARM and SpiNNaker architecture, Manchester University). As part of this project, I will schedule meetings with these individuals to develop partnerships for future joint research. However, we are also open to new collaborations with the ICT industry. For example, the School of Computing Science has existing links with Microsoft and BT that might lead to further applications in the ICT industry. Next steps could be a CASE PhD studentship with these companies.
A specific application is to use the knowledge of design principles of human brains to improve the design of the SpiNNaker neurochip developed in the UK. Currently, communication between nodes is a potential bottleneck for larger numbers of network nodes. However, the hierarchical modular organisation of the brain shows principles that reduce the number of processing steps and the amount of energy that is needed (Kaiser & Hilgetag, PLoS Computational Biology, 2006). This collaboration might therefore breach the current limits for designing large-scale artificial information processing systems.
Linking brain connectivity studies with ICT applications is also evident in my previous work: comparing the architecture of computers and brains (Kaiser, Phil Trans Roy Soc A, 2007), transferring network analysis routines to predictions of drug targets (EPSRC CASE studentship with e-Therapeutics), and using ideas from brain organization on wireless communication networks (Brust et al., IEEE Intl Conf. on Communication, 2012).
Results will, wherever possible, be published in open-access journals. In addition, reprints will be available on my website (www.biological-networks.org/) and within the arxiv.org pre-print server. Software developed through this project will be available on my website but also in standard repositories of the field (NITRC, INCF Software database, SourceFourge). I will also disseminate results through national and international meetings. Reaching both industry and academia, I will use mailing lists and meetings of the UK INCF Neuroinformatics node to present outcomes of this research.
As in the past with BBC Radio and U.S. News, I will reach general audiences through press releases, newspaper reports, and radio interviews. I will also give public lectures about the results, for example within Newcastle SciScreen and the 2013 British Science Association Festival at Newcastle. I have attended workshops on public engagement and media relations in the past and will work more actively with the media to reach the general public.
This project might also inform the design of technical communication and information processing systems. We will therefore discuss applications with our partners in the IT industry. In particular, we will use our existing links with Dr Kozloski (IBM T.J. Watson Research Center, New York) and Prof Steve Furber (developer of ARM and SpiNNaker architecture, Manchester University). As part of this project, I will schedule meetings with these individuals to develop partnerships for future joint research. However, we are also open to new collaborations with the ICT industry. For example, the School of Computing Science has existing links with Microsoft and BT that might lead to further applications in the ICT industry. Next steps could be a CASE PhD studentship with these companies.
A specific application is to use the knowledge of design principles of human brains to improve the design of the SpiNNaker neurochip developed in the UK. Currently, communication between nodes is a potential bottleneck for larger numbers of network nodes. However, the hierarchical modular organisation of the brain shows principles that reduce the number of processing steps and the amount of energy that is needed (Kaiser & Hilgetag, PLoS Computational Biology, 2006). This collaboration might therefore breach the current limits for designing large-scale artificial information processing systems.
Linking brain connectivity studies with ICT applications is also evident in my previous work: comparing the architecture of computers and brains (Kaiser, Phil Trans Roy Soc A, 2007), transferring network analysis routines to predictions of drug targets (EPSRC CASE studentship with e-Therapeutics), and using ideas from brain organization on wireless communication networks (Brust et al., IEEE Intl Conf. on Communication, 2012).
Results will, wherever possible, be published in open-access journals. In addition, reprints will be available on my website (www.biological-networks.org/) and within the arxiv.org pre-print server. Software developed through this project will be available on my website but also in standard repositories of the field (NITRC, INCF Software database, SourceFourge). I will also disseminate results through national and international meetings. Reaching both industry and academia, I will use mailing lists and meetings of the UK INCF Neuroinformatics node to present outcomes of this research.
As in the past with BBC Radio and U.S. News, I will reach general audiences through press releases, newspaper reports, and radio interviews. I will also give public lectures about the results, for example within Newcastle SciScreen and the 2013 British Science Association Festival at Newcastle. I have attended workshops on public engagement and media relations in the past and will work more actively with the media to reach the general public.
Organisations
- Newcastle University (Lead Research Organisation)
- Intel (United States) (Collaboration)
- European Organization for Nuclear Research (CERN) (Collaboration)
- University of Manchester (Project Partner)
- University of Nottingham (Project Partner)
- Sree Chitra Thirunal Institute for Medical Sciences and Technology (Project Partner)
- University of Bonn (Project Partner)
Publications
Ainsworth M
(2016)
GABAB receptor-mediated, layer-specific synaptic plasticity reorganizes gamma-frequency neocortical response to stimulation.
in Proceedings of the National Academy of Sciences of the United States of America
Bauer R
(2017)
Nonlinear growth: an origin of hub organization in complex networks.
in Royal Society open science
Bauer R
(2021)
Creative Destruction: A Basic Computational Model of Cortical Layer Formation.
in Cerebral cortex (New York, N.Y. : 1991)
Bauer R
(2017)
Brain Evolution by Design
Cromarty RA
(2018)
Structural Brain Correlates of Attention Dysfunction in Lewy Body Dementias and Alzheimer's Disease.
in Frontiers in aging neuroscience
Edwin Thanarajah S
(2016)
Abnormal Connectional Fingerprint in Schizophrenia: A Novel Network Analysis of Diffusion Tensor Imaging Data.
in Frontiers in psychiatry
Gonzalez-De-Aledo P
(2018)
An optimization approach for agent-based computational models of biological development
in Advances in Engineering Software
| Description | Girls connectome matures earlier: selective pruning stabilizes network despite massive changes Newcastle University scientists have discovered that as the brain re-organises connections throughout our life, the process begins earlier in girls which may explain why they mature faster during the teenage years. As we grow older, our brains undergo a major reorganisation reducing the connections in the brain. Studying people up to the age of 40, scientists led by Dr Marcus Kaiser and Ms Sol Lim at Newcastle University as part of the Human Green Brain project found that while overall connections in the brain get streamlined, long-distance connections that are crucial for integrating information are preserved. The researchers suspect this newly-discovered selective process might explain why brain function does not deteriorate - and indeed improves - during this pruning of the network. Interestingly, they also found that these changes occurred earlier in females than in males. Explaining the work which is being published in Cerebral Cortex, Dr Kaiser, Associate Professor in Neuroinformatics at Newcastle University, says: "Long-distance connections are difficult to establish and maintain but are crucial for fast and efficient processing. If you think about a social network, nearby friends might give you very similar information - you might hear the same news from different people. People from different cities or countries are more likely to give you novel information. In the same way, some information flow within a brain module might be redundant whereas information from other modules, say integrating the optical information about a face with the acoustic information of a voice is vital in making sense of the outside world." The researchers at Newcastle, Glasgow, and Seoul Universities evaluated the scans of 121 healthy participants between the ages of 4 and 40 years as this is where the major connectivity changes can be seen. Using a non-invasive technique called diffusion tensor imaging - a special measurement protocol for Magnetic Resonance Imaging (MRI) scanners - they demonstrated that fibres are overall getting pruned that period. However, they found that not all projections (long-range connections) between brain regions are affected to the same extent; changes were influenced differently depending on the types of connections. These projections are short-cuts that quickly link different processing modules, e.g. for vision and sound, and allow fast information transfer and synchronous processing. Changes in these connections have been found in many developmental brain disorders including autism, epilepsy and schizophrenia. The researchers have demonstrated for the first time that the loss of white matter fibres between brain regions is a highly selective process - a phenomenon they call preferential detachment. They show that connections between distant brain regions, between brain hemispheres, and between processing modules lose fewer nerve fibres during brain maturation than expected. The researchers say this may explain how we retain a stable brain network during brain maturation. Commenting on the fact that these changes occurred earlier in females than males, Ms Sol Lim explains: "The loss of connectivity during brain development can actually help to improve brain function by reorganizing the network more efficiently. Say instead of talking to many people at random, asking a couple of people who have lived in the area for a long time is the most efficient way to know your way. In a similar way, reducing some projections in the brain helps to focus on essential information." Academic paper: Preferential Detachment During Human Brain Development: Age- and Sex-Specific Structural Connectivity in Diffusion Tensor Imaging (DTI) Data, Sol Lim, Cheol Han, Peter Uhlhaas and Marcus Kaiser. Cerebral Cortex. |
| Exploitation Route | Results about the self-organisation of neural systems can inform the understanding of developmental diseases and the implementation of artificial neural networks. |
| Sectors | Digital/Communication/Information Technologies (including Software),Healthcare |
| URL | http://cercor.oxfordjournals.org/content/early/2013/12/13/cercor.bht333.full |
| Description | Research from this study has led to the BioDynamo software framework (https://biodynamo.org/ ) for simulating biological tissue development. Since 2016, we have grown BioDynamo to a consortium that includes 8 major institutions (University of Surrey, Newcastle University, GSI Helmholtz Centre, SCImPULSE Foundation, Immunobrain Checkpoint, University of Geneva, University of Cyprus) from six different countries (UK, Switzerland, Germany, Netherlands, Israel and Cyprus). All consortium members have signed a collaboration agreement, and we have regular bi-weekly meetings as well as consortium meetings every 6 months. We also hold regular events (e.g., in November 2021 we held the BioDynaMo workshop on agent-based modelling with over 70 international registrations) and have plans to create a legal entity that provides training workshops and education in Computational Biomedicine. This has led to further impact and funding including modelling Covid-19: Innovation Fellowship: Computational modelling of cryopreservation of biological tissue Funded value: £502.399 Project reference: EP/S001433/1 Computational modelling of retinal development Funded value: £219.581 Project reference: MR/N015037/1 BioDynaMo agent-based simulations for COVID-19 Funded value: CHF80.000 Project reference: Covid-19 Fast-Track Funding |
| First Year Of Impact | 2016 |
| Sector | Digital/Communication/Information Technologies (including Software),Healthcare,Pharmaceuticals and Medical Biotechnology |
| Impact Types | Societal,Economic |
| Title | BioDynamo |
| Description | An open-source software platform to easily create, run, and visualise 3D agent-based simulations. Originally BioDynaMo was developed to model biological dynamics in various contexts, like simulating a wide range of biological processes, such as cell proliferation, neurite growth and electrical activity, by exploiting modern computer systems in a very efficient manner. As complex biological systems typically consists of many million to hundreds of millions of cells the BioDynaMo computing core needed to be very efficient and make optimially use of the many available cores and GPU's in typical modern computing hardware. For this reason BioDynaMo is also written in modern C++ to be as close to the hardware as possible. Thanks to this relentless focus on efficiency large biological systems can be simulated in acceptable times and BioDynaMo performs extremely well when modelling systems with less than a few million agents. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2016 |
| Provided To Others? | Yes |
| Impact | Since 2016, we have grown the collaboration to a consortium that includes 8 major institutions (University of Surrey, Newcastle University, GSI Helmholtz Centre, SCImPULSE Foundation, Immunobrain Checkpoint, University of Geneva, University of Cyprus) from six different countries (UK, Switzerland, Germany, Netherlands, Israel and Cyprus). All consortium members have signed a collaboration agreement, and we have regular bi-weekly meetings as well as consortium meetings every 6 months. We also hold regular events (e.g., in November 2021 we held the BioDynaMo workshop on agent-based modelling with over 70 international registrations) and have plans to create a legal entity that provides training workshops and education in Computational Biomedicine. |
| URL | https://biodynamo.org |
| Description | CERN Openlab |
| Organisation | European Organization for Nuclear Research (CERN) |
| Department | CERN - Other |
| Country | Switzerland |
| Sector | Academic/University |
| PI Contribution | Supervision of students at CERN. |
| Collaborator Contribution | Sponsorship of internship student (3 months) and technical student (six months) who was working on our project but was based at CERN. |
| Impact | Software development. |
| Start Year | 2015 |
| Description | Intel |
| Organisation | Intel Corporation |
| Country | United States |
| Sector | Private |
| PI Contribution | Provision of computer program that needed to be optimized as part of the Intel student competition. |
| Collaborator Contribution | Hosting of the Intel Code Modernization Challenge with more than 2,000 registered students. |
| Impact | A conference paper about the parallelization and optimization approaches for the winning code is in preparation. |
| Start Year | 2015 |
| Title | BioDynamo |
| Description | An open-source software platform to easily create, run, and visualise 3D agent-based simulations. Originally BioDynaMo was developed to model biological dynamics in various contexts, like simulating a wide range of biological processes, such as cell proliferation, neurite growth and electrical activity, by exploiting modern computer systems in a very efficient manner. As complex biological systems typically consists of many million to hundreds of millions of cells the BioDynaMo computing core needed to be very efficient and make optimially use of the many available cores and GPU's in typical modern computing hardware. For this reason BioDynaMo is also written in modern C++ to be as close to the hardware as possible. Thanks to this relentless focus on efficiency large biological systems can be simulated in acceptable times and BioDynaMo performs extremely well when modelling systems with less than a few million agents. |
| Type Of Technology | Software |
| Year Produced | 2016 |
| Open Source License? | Yes |
| Impact | Since 2016, we have grown the collaboration to a consortium that includes 8 major institutions (University of Surrey, Newcastle University, GSI Helmholtz Centre, SCImPULSE Foundation, Immunobrain Checkpoint, University of Geneva, University of Cyprus) from six different countries (UK, Switzerland, Germany, Netherlands, Israel and Cyprus). All consortium members have signed a collaboration agreement, and we have regular bi-weekly meetings as well as consortium meetings every 6 months. We also hold regular events (e.g., in November 2021 we held the BioDynaMo workshop on agent-based modelling with over 70 international registrations) and have plans to create a legal entity that provides training workshops and education in Computational Biomedicine. |
| URL | https://biodynamo.org |