Advanced FMRI acquisition, reconstruction and signal processing for dynamic brain network imaging

Lead Research Organisation: University of Oxford
Department Name: Clinical Neurosciences


How is the brain wired? Are all of our brains wired the same way? What happens when the brain's connections fail? For the first time, scientists are beginning to be able to ask these questions in living human subjects using non-invasive brain imaging methods like functional MRI (FMRI). FMRI measures brain activity; for example, it can show you what part of the brain responds when you look at a picture of a face. Interestingly, one of the most powerful ways to study brain connections is with FMRI of the brain "at rest", when the subject hasn't been asked to perform any specific task. In this state, all of the brain's networks can be observed to be "talking" amongst themselves simultaneously. If we observe this "random chatter" for long enough, we can determine which brain regions are connected to each other because they will tend to be active at the same time. These "resting state networks" have been shown to be the very same networks that are involved in active thought processes and mental tasks, and they are altered in neurological disease. They therefore offer a very powerful window into brain function and health.

Currently, our ability to study these brain networks is severely limited by the quality of FMRI data and sophistication of analysis to which scientists have access. Our project aims to improve both of these aspects of resting FMRI experiments. To do this, we have to address several tricky engineering challenges.

The first goal of our project is to improve the FMRI data quality. One major improvement that would enable us to study the brain's dynamics with greater accuracy is to acquire images faster. Conventional MRI acquires one data point per image pixel, and the more pixels an image has, the longer it takes to acquire. Since current techniques already operate at the "speed limit" of conventional MRI, we have to take a different approach to the problem. We will take advantage of the same mathematical principles that underlie image and video compression on the internet: sometimes images can be described using fewer information "bits" than the total number of pixels in the image. So images that are "compressible" can be assembled from far less data, enabling faster image acquisition. Our challenge is to find the best way to compress FMRI images given the properties of our resting networks. As part of this, we will take advantage of the state-of-the-art in MRI scanners, operating with 2-4 times stronger magnetic field than is currently common in hospitals.

The second goal of our project is to take advantage of this improved data to obtain better measures of brain connections. Currently, we can use resting FMRI to create images of areas that are connected to each other, but we don't know how these regions are connected. Just as one cannot plan a route from an atlas that shows cities but no roads, these brain maps cannot really tell us anything about how information flows in the brain. The first step, then, is to establish which brain regions have direct connections (roads) between them. In the brain, connections are often "one way", so the second step is to be able to detect the direction of connections. We will develop statistical tools for robustly finding these network properties, which will enable us to establish the general flow of information in the brain. However, many of the brain's most interesting properties come from the "dynamics" of these networks: connections may become stronger or weaker over time, or could shut down temporarily. For example, this is thought to be important in our ability to shift our attention between different tasks or thoughts. Our final goal, then, is to be able to detect the dynamics of both individual connections and interacting networks. For this, we will develop sophisticated methods for categorizing network states that may be changing over time.

These methods will enable neuroscientists to study the brain at an unprecedented level of detail.

Planned Impact

Our proposal is expected to have major impact in a range of spheres outside of academia, including Society, Economy and Knowledge.

Impact on Society will largely be through improvements in healthcare and quality of life. The resting-state FMRI methods developed here have enormous potential as a tool for early detection of neurological disorders. For example, there is increasing evidence that resting brain fluctuations reflect the full range of the brain's functional capabilities. This raises the possibility of using resting FMRI to discover residual brain function that might respond to treatment in patients who cannot perform a task particularly well (e.g., Alzheimer's patients with memory deficits or stroke patients with motor impairment). In addition, our MRI acquisition methods could be used with a specific task in epilepsy or tumour pre-surgical planning. Resting-state FMRI also has tremendous potential as a marker of changes in the brain's connectivity, for use in population-wide studies. This would be particularly helpful to advance our understanding of diseases of connectivity such as autism, schizophrenia and epilepsy. There is already evidence that resting FMRI can detect some changes in connectivity long before onset of symptoms, for example in individuals with a genetic risk factor for Alzheimer's. There is already interest from the pharmaceutical industry in exploring the potential of resting FMRI in drug development, and we also envision uses in evaluating other forms of therapy (e.g., devices or rehabilitation regimens). The ability to detect the early changes has an enormous impact on our ability to identify the most promising therapies, improving health care and quality of life.

Impact on Economy is also tied to healthcare and quality of life. Improved diagnosis and treatment of disease can lead to tremendous financial savings in the healthcare sector where the economic burden of an aging population is an increasing problem. Any possible improvement in either patient diagnosis or population studies carries an economic benefit. There is also economic impact associated with the techniques themselves within the global market in healthcare. For example, the world-wide market for MRI scanners was $4.4 billion in 2010, with just three vendors dominating the market (GE, Siemens and Philips). Manufacture of the superconducting magnets that form the basis of these systems is a major industry in the UK. Similarly, drug trials run by pharmaceutical companies cost millions of pounds, and several UK companies have been founded to analyze data (including images) for these studies.

Impact on Knowledge will occur from our more general understanding of how the brain works. Resting FMRI measures the brain's intrinsic oscillations, the purpose and importance of which remain controversial and mysterious. Some scientists place them at the very heart of brain function (G. Buzsaki, "Rhythms of the Brain"). In the context of our project, these oscillations are known to reflect the brain's functional connections. Methods for mapping these connections in living humans are a recent technological advance that has precipitated a world-wide effort to provide a map of the brain's connections. Our group is part of a leading consortium, the US-based Human Connectome Project, which aims to provide the most comprehensive and rich in vivo map of the brain's connections. Our research will contribute to our understanding of the brain, helping to answer of the great remaining scientific questions: how does the brain work?


10 25 50
Description A grand scientific challenge for the coming decade is to understand how connections in the brain relate to brain function and health. We have developed new methods for acquiring and analysing magnetic resonance imaging data to enable neuroscientists to ask these questions. These methods provide improved sensitivity to detecting subtle aspects of brain connections. Using data from the Human Connectome Project, we demonstrated novel insights into the brain based on connectivity. In particular, we identified patterns of connections that track a broad range of lifestyle, demographic and health factors, which appear to indicate a relationship between brain connectivity and overall 'life fitness'.
Exploitation Route Our methods can be adopted by basic neuroscientists in academia, by industrial researchers in pharmaceutical and therapeutics and ultimately by practicing clinicians. In particular analysis and acquisition techniques are available for sharing. We have a track record of sharing acquisition techniques in collaboration with scanner vendors. Analysis methods are shared through the popular FMRIB Software Library (FSL).
Sectors Healthcare,Pharmaceuticals and Medical Biotechnology

Description Coleman-Cohen Fellowship
Amount £10,000 (GBP)
Organisation British Technion Society 
Sector Charity/Non Profit
Country United Kingdom
Start 09/2015 
End 08/2016
Description Royal Academy of Engineering Fellowship
Amount £49,971,538 (GBP)
Funding ID RF201617\16\23 
Organisation Royal Academy of Engineering 
Sector Learned Society
Country United Kingdom
Start 09/2017 
End 08/2022
Description Wellcome Trust Senior Research Fellowship
Amount £1,793,980 (GBP)
Organisation Wellcome Trust 
Sector Charity/Non Profit
Country United Kingdom
Start 10/2016 
End 09/2021
Title FSL 
Description FSL is a comprehensive software package for neuroimaging analysis that covers all of the main MRI neuroimaging modalities (task-functional, resting-functional, structural, diffusion, perfusion). 
IP Reference  
Protection Protection not required
Year Protection Granted
Licensed Yes
Impact The FMRIB Software Library (FSL) is one of the two most widely used software packages for neuroimaging research and clinical applications of neuroimaging in the world, used by over 1000 universities and hospitals worldwide and downloaded 160,000 times in the last five years. It is freely distributed and supported for non-commercial use, making state-of-the-art analysis tools available to the global academic community. The three primary publications documenting FSL have been cited over 10,000 times. FLS is also licensed commercially through Oxford University Innovation, and is currently licensed by 7 of the top 10 pharmaceutial companies. Commerical licensing has generated over £1m of income.
Title Low Rank Matrix Completion for FMRI 
Description In functional MRI (fMRI), faster sampling of data can provide richer temporal information and increase temporal degrees of freedom. However, acceleration is generally performed on a volume-by-volume basis, without consideration of the intrinsic spatio-temporal data structure. We present a novel method for accelerating fMRI data acquisition, k-t FASTER (FMRI Accelerated in Space-time via Truncation of Effective Rank), which exploits the low-rank structure of fMRI data. 
Type Of Technology New/Improved Technique/Technology 
Year Produced 2014 
Impact facilitating other research by accelerating fMRI data acquisition 
Title Motion correction for functional MRI with three-dimensional hybrid radial-Cartesian EPI 
Description Subject motion is a major source of image degradation for functional MRI (fMRI), especially when using multishot sequences like three-dimensional (3D EPI). We present a hybrid radial-Cartesian 3D EPI trajectory enabling motion correction in k-space for functional MRI. 
Type Of Technology New/Improved Technique/Technology 
Year Produced 2016 
Impact Facilitation of other research by improvement of fMRI data through motion artefact reduction, especially for subject groups with significant head motion. 
Title Non-Cartesian FMRI with Low-Rank Constraints and Sensitivity Encoding 
Description This work generalises the k-t FASTER approach for rank-constrained FMRI data reconstruction by permitting more general linear encoding of the imaging data, including non-Cartesian (e.g. radial) sampling trajectories and coil sensitivity encoding. 
Type Of Technology New/Improved Technique/Technology 
Year Produced 2016 
Impact facilitation of other research by accelerating fMRI data acquisition 
Description We bring together leading neuro-imaging experts to provide our clients with the best solutions for their research questions. Together with the client's study team, we enable functional and structural MRI to fulfil its rich potential as a powerful, non-invasive biomarker for a wide range of diseases and disease treatments. We determine the optimal processing pipeline for the client's data, in collaboration with the clinical trial team and the affiliated CRO. This allows us to tailor our services to a client's specific needs, implementing those steps that are regarded state-of-the-art in the neuroimaging domain that the client targets. SBGneuro provides these expert services with a maximum of security and a minimum of overhead. Our analysis pipeline begins with the receipt of de-identified data from the CRO. We offer industry-standard mechanisms for secure and encrypted data upload onto dedicated servers, followed by local quality control, optimised data pre-processing, and structural and functional analysis. We extract a large repertoire of (f)MRI outcome measures, each representing a valid and distinct biomarker of activity or connectivity. At study completion or upon completion of interim analysis we deliver results to the CRO or directly to the clinical trial team. We thereby adhere to specified data transfer specifications or deliver fully customized imaging analyses reports. This stream-lined model allows us to deliver secure, robust results in a timely, flexible, and cost-efficient manner. Our post-trial services provide guidance on data interpretation and assistance with publication of results. 
Year Established 2012 
Impact The company has produced ontracts with pharmaceutial research, including CROs, totalling at least £3m to date.
Description Article in Scientific American about Biobank 
Form Of Engagement Activity A press release, press conference or response to a media enquiry/interview
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Public/other audiences
Results and Impact Article in Scientific American about the first results of the Biobank study, including an interview with me.
Year(s) Of Engagement Activity 2016
Description Brain Awareness week panel 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Public/other audiences
Results and Impact A free Brain Awareness Week screening of a documentary about renowned academic and research scientist Dr Marian Diamond, organised by the Nuffield Department of Clinical Neurosciences. I participated in a panel discussion with other Oxford University neuroscientists following the film.
Year(s) Of Engagement Activity 2018
Description Curiosity Carnival 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Public/other audiences
Results and Impact The Curiosity Carnival on Friday 29 September was a chance to find out what research is really all about, meet researchers, ask questions and discover how research affects and changes all our lives.

The night was a huge festival of curiosity - a city-wide programme of activities across the University of Oxford's museums, libraries, gardens and woods. There was a wide range of activities for all ages and interests - live experiments, games, stalls, busking, debates, music, dance and a pub-style quiz.

Oxford's Curiosity Carnival 2017 joined hundreds of other European cities in celebrating European Researchers' Night.
Year(s) Of Engagement Activity 2018
Description Interview on BBC Radio 4's 'All in the Mind' 
Form Of Engagement Activity A press release, press conference or response to a media enquiry/interview
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Public/other audiences
Results and Impact Interview for the Radio 4 programme 'All in the Mind'
Year(s) Of Engagement Activity 2015