A Computational Prototype for Electroencephalographic Brain Connectomics

Lead Research Organisation: University of Kent
Department Name: Sch of Computing

Abstract

What are the neural signatures of consciousness? This elusive yet fascinating challenge for neuroscience takes on an immediate clinical and societal significance in patients diagnosed to be in vegetative and minimally conscious states, collectively termed disorders of consciousness (DoC). In recent years, rapid advancements in the study of human brain connectomics have generated valuable insights into how networks of interactions between specific brain regions support consciousness, and underlie neurological and psychiatric disorders. This research has the potential to improve diagnosis and prognosis for DoC patients, with significant ethical and healthcare implications. To realise this potential, this project will develop and validate robust brain connectomics software for assessing consciousness in patients at the bedside.

Despite significant advances in the neuroscience of consciousness, there are no brain-based tests of conscious state available clinically for diagnostic and prognostic applications in DoC. Current guidance in clinical practice relies primarily on behavioural assessment to characterise the level of consciousness, which has been shown to incur a risk of misdiagnosis as high as 40%. The use of brain-based measures has yet to find implementation and acceptance in clinical practice, and in an individual patient's rehabilitation journey in particular.

This proposal aims to directly contribute to addressing this serious limitation, by creating a software prototype that bring together state of the art computational analysis of brain activity data. This data has already been acquired right at the patient's bedside using electroencephalography (EEG), a non-invasive, portable brain mapping technology. This prototype will incorporate a computational data analytics pipeline designed to enable detailed visualisation and quantification of brain connectivity networks in DoC patients. These networks will be estimated by applying sophisticated signal processing methods to data collected from patients by clinical partners. We will develop and validate metrics derived from graph theory to characterise the topological and topographical properties of brain networks measured with these data. Further, the pipeline will apply machine learning to automatically classify the state of consciousness in individual patients based on the signatures in their brain networks. Most importantly, the software prototype will be designed in such a way that it can be easily deployed at the bedside in clinical rehabilitation centres where DoC patients are resident.

If realised, this project will be one of the first to directly contribute to the scientific advancement and validation of healthcare technology with real-world impact in the assessment of consciousness in the clinic. It will be of particular interest to patients' families and carers, in addition to medical professionals involved in their care, treatment and management. It has the potential to enable them to have access to objective, discriminative, and cost-effective information about brain activity related to consciousness, available right at the bedside. This will ultimately facilitate informed decision-making on behalf of patients, and the employment of more effective therapeutic interventions. This advance will eventually also have broader implications for the ethical and legal debate in the public surrounding the assessment of consciousness in the clinic.

Planned Impact

Ultimately, the primary beneficiaries of this research will be patients diagnosed with disorders of consciousness (including the vegetative and minimally conscious states). According to recent estimates, 4000-16000 patients in the UK have a diagnosis of vegetative state, with three times as many in minimally conscious states. This poses a serious clinical challenge, as the level of consciousness in as many as 40% of patients could be misdiagnosed. The central aim of this proposal is the development of a novel prototype for characterising brain networks in patients. Once the proposed software prototype has been developed, tested and validated, we aim to pilot its deployment it at the patient's bedside at both centres. Eventually, through the development of a turnkey product or service developed for clinical end users, the results of such assessments will contribute to more accurate diagnosis of consciousness, and better prognostication of recovery. We will enable this by working with clinicians to define best practice for incorporating brain-based assessments into clinical decision making in individual cases.

Clinical professionals, including neurorehabilitation experts and allied health professionals working with DoC patients will hence benefit from access to the outcomes of this research. These outcomes will not only complement and augment existing clinical tools, they will enable clinicians to more accurately track the variable response to the specific pharmacological and therapeutic interventions they trial with individual patients. Ultimately, the availability of EEG-based brain imaging assessments at the bedside will enable clinicians to adopt a precision medicine approach toward more accurate patient stratification. This will ultimately enable them to provide patients with improved access to tailored care and therapy that best meets the needs of each individual.

The PI has contributed to increased interaction between science and policy making, by providing input to a recent Parliament Office of Science and Technology document on prolonged DoC. He also contributed to a special event on diagnostic and prognostic challenges, hosted at the Houses of Parliament, bringing together academics, clinicians, members of the judiciary, policy makers, patients' families and carers. As the proposed brain network analysis software reaches maturity, the PI will inform and disseminate knowledge about it at similar public forums in the future. He will also make contact, interact with and disseminate knowledge within the clinical bodies involved in policy making and guideline development in this context.

This proposal will also have indirect impact for patients in a related and equally complex clinical context: monitoring consciousness during general anaesthesia. The risks of intra-operative awareness and consequent post-operative delirium, which are particularly challenging in some patient groups, could be minimised with advances in brain monitoring with EEG. The work here on developing robust computational tools to track consciousness at the bedside will also influence future academic and clinical research and development into viable solutions for monitoring consciousness in the operating theatre.

More broadly, this research will also have impact amongst the general public, and students in particular. Acute and chronic disorders of consciousness have attracted considerable public interest, due to the complex ethical, legal and moral questions it raises for a caring society. Over the course of the proposed research, the PI will engage with the ongoing public debate in this context. This will include at least two public talks each year during the lifetime of this grant, at science events and festivals. At these forums, our research outcomes will demonstrate current advances in translational and personalised medicine, and the showcase the potential for novel healthcare technologies to inform the debate.

Publications

10 25 50
 
Description We have developed a software prototype that brings together state of the art computational analysis of brain activity to benefit patients with severe brain injury at their bedside. This prototype includes algorithms to visualise and measure networks of brain activity in such patients. These outputs are designed to enable clinicians better diagnose the state of awareness in patients, and prognosticate recovery from disorders of consciousness.

Networks of brain activity are estimated by applying sophisticated signal processing methods to electrical activity data collected from patients' brains at their bedside. We have developed and validated metrics to characterise the networks measured with these data. Further, our pipeline applies machine learning to automatically classify the state of consciousness in individual patients based on their brain networks. Most importantly, the software prototype has been designed in such a way that it can be deployed along with high-density EEG systems at the bedside in rehabilitation centres where patients are resident.
Exploitation Route The software pipeline we have developed can serve a valuable role in translating neurotechnology to make it available at the patient's bedside. In doing so, it will address an unmet need in the care, treatment and management of patients with disorders of consciousness, which includes the so-called vegetative and minimally conscious states. As a next step, the research and development we have conducted could be taken forward to conduct a larger-scale technology trial of this software as a medical device, to obtain the regulatory approvals necessary to make it available more widely in the UK National Health Service.

In the longer term, the development and validation of such technologies also aim to update current clinical guidelines for the care of patients. The latest version of these guidelines, issued by the Royal College of Physicians, highlights the current lack of such validation of brain-based methods to complement behavioural assessments of consciousness after severe brain injury.
Sectors Digital/Communication/Information Technologies (including Software)

Healthcare

URL https://www.github.com/srivaschennu/MOHAWK/wiki
 
Description Clinical engagement and deployment - The software prototype developed here has been deployed at three clinical centres for automatic diagnosis of consciousness in patients after severe brain injury. They are located in the UK (Addenbrooke's Hospital, Cambridge), Belgium (University Hospital, Liege), and Germany (Therapiezentrum Burgau, Munich). All three are major national centres treating and managing patients with chronic disorders of consciousness following such injury. The PI has travelled to the centres to deploy the software prototype and trained clinical teams in its use for generating diagnostic outputs for clinical reporting. As of date, the software has been used to assess 100+ patients. Short film about this research and its impact - The PI contributed to the production of a short film titled "The Mohawk of Consciousness" broadcast on KMTV. This film has a specific regional focus and includes interviews with patients and families. It highlights how the healthcare technology we are developing could help improve diagnosis, prognosis and treatment of patients with brain injury. Since the deployment of the software, data from 100+ patients have been analysed using it. The reports generated from our software have been included in reports provided to patients' families and treating clinicians. Further, the software is also being used alongside therapeutic interventions like transcranial direct current stimulation (tDCS), to quantify and measure the response to therapy. As a result of this project and others in this space, there is now wider awareness of the value of bedside neuroimaging for assessment, diagnosis and treatment of patients with disorders of consciousness following brain injury.
First Year Of Impact 2021
Sector Healthcare
Impact Types Cultural

Societal

 
Description An enhanced online graphics processing unit facility for Early Career Researchers
Amount £99,620 (GBP)
Funding ID EP/S017755/1 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Public
Country United Kingdom
Start 08/2018 
End 03/2020
 
Description REF impact case study development award
Amount £1,200 (GBP)
Organisation University of Kent 
Sector Academic/University
Country United Kingdom
Start 11/2018 
End 07/2019
 
Title The MOHAWK software toolbox for brain network analysis and visualisation 
Description We have developed a software prototype that uses employs computational analysis of brain activity to diagnose consciousness in patients with severe brain injury, right at their bedside. This prototype employs algorithms to automatically visualise and measure networks of brain activity in patients. These outputs generated by the software are designed to enable clinicians better diagnose the state of awareness in patients, and prognosticate recovery from disorders of consciousness. These networks are estimated by applying sophisticated signal processing methods to electrical activity data collected from patients' brains at their bedside, in collaboration with clinical partners. We have developed and validated metrics to characterise the networks measured with these data. Further, our pipeline applies machine learning to automatically classify the state of consciousness in individual patients based on their brain networks. Most importantly, the software prototype has been designed in such a way that it can be deployed along with high-density EEG systems at the bedside in rehabilitation centres where patients are resident. 
Type Of Material Physiological assessment or outcome measure 
Year Produced 2018 
Provided To Others? Yes  
Impact Other research groups and industrial partners are using this software toolbox, both for novel scientific research and for training new researchers and clinicians. Specifically: 1. Researchers at the Universities of Cambridge, Birmingham and Liege (Belgium) are using and building upon the software we have provided under an open-access GNU GPL. They are using it to conduct novel research independent of this research study. 2. Philips Neuro is using the software to introduce brain connectivity analysis to users of their high-density hardware. This is taking place during regular training workshops that Philips Neuro organises for their existing and potential customers, often attached to relevant scientific/clinical conferences, at locations around the world. 
URL https://github.com/srivaschennu/MOHAWK
 
Title Research data supporting "Transient Topographical Dynamics of the Electroencephalogram Predict Brain Connectivity and Behavioural Responsiveness During Drowsiness" 
Description This set contains 19 EEG recordings of different participants performing an auditory semantic task as they fell asleep. The participants were directed to not consume stimulants like coffee and to sleep 1-2 h less than normally before the experiment. All of the subjects were assessed as easy sleepers on the Epworth Sleepiness Scale (scores 7-14). The stimuli consisted of 96 spoken English words chosen from the CELEX lexical database. Half of the words denoted animals and the other half denoted objects. The subjects were asked to classify each stimulus in its respective category (animal or object) by pressing a button. The stimuli were presented through headphones, with an average distance of 8.4 s (minimum 6.2 s) between consecutive stimuli, as the subjects were lying with their eyes closed in a reclining chair. To facilitate drowsiness, the task was performed in a dark, acoustically and electrically shielded EEG room, and the participants were told that they could fall asleep at any point during the experiment, although they were asked not to stop responding deliberately while still awake. The electroencephalogram was continuously recorded at 500 samples per second from 64 Ag/AgCl electrodes positioned and labelled according to the extended 10/20 system, with Cz as a reference and including vertical and horizontal electrooculography channels. 
Type Of Material Database/Collection of data 
Year Produced 2018 
Provided To Others? Yes  
URL https://www.repository.cam.ac.uk/handle/1810/286670
 
Description Addenbrooke's Hospital, Cambridge 
Organisation Addenbrooke's Hospital
Department Department of Medicine
Country United Kingdom 
Sector Hospitals 
PI Contribution The software prototype we have developed has been deployed at the partner site.
Collaborator Contribution This partner contributed part of the initial patient dataset to help build the prototype, comprising behavioural and high-density EEG measurements. They also provided valuable clinical input, guidance and support that ensured the success of the overall project.
Impact This collaboration has led to joint peer-reviewed publications in high impact journals (DOIs: 10.1093/brain/awx163, 10.3389/fneur.2018.00676).
Start Year 2017
 
Description Therapiezentrum Burgau 
Organisation Therapy Center Burgau
Country Germany 
Sector Hospitals 
PI Contribution The software prototype we have developed has been deployed at the partner site.
Collaborator Contribution This partner contributed part of the initial patient dataset to help build the prototype, comprising behavioural and high-density EEG measurements.
Impact This collaboration has led to clinical impact Therapy Center Burgau, where patients have been assessed using the brain network assessment software we have supplied, and feedback on the findings therefrom have been provided to families and clinicians.
Start Year 2018
 
Description University Hospital of Liege 
Organisation Central University Hospital of Liege
Country Belgium 
Sector Hospitals 
PI Contribution The software prototype we have developed has been deployed at the partner site.
Collaborator Contribution This partner contributed part of the initial patient dataset to help build the prototype, comprising behavioural and high-density EEG measurements. They also provided valuable clinical input, guidance and support that ensured the success of the overall project.
Impact This collaboration has led to joint peer-reviewed publications in high impact journals (DOIs: 10.1093/brain/awx163, 10.1016/j.brs.2018.09.002).
Start Year 2017
 
Title MOHAWK 
Description MOHAWK is a software prototype that brings together state of the art computational analysis of brain activity to benefit patients with severe brain injury at their bedside. This prototype includes algorithms to visualise and measure networks of brain activity in such patients. These outputs are designed to enable clinicians better diagnose the state of awareness in patients, and prognosticate recovery from disorders of consciousness. 
Type Of Technology Software 
Year Produced 2019 
Open Source License? Yes  
Impact The MOHAWK software package has been used to assess the level of consciousness in patients after severe brain injury at three clinical centres in Europe. They are located in the UK (Addenbrooke's Hospital, Cambridge), Belgium (University Hospital, Liege), and Germany (Therapiezentrum Burgau, Munich). 
URL https://github.com/srivaschennu/MOHAWK/wiki
 
Description Article for The Conversation 
Form Of Engagement Activity Engagement focused website, blog or social media channel
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Public/other audiences
Results and Impact The PI wrote an article for The Conversation, highlighting the challenges and potential for the application of machine learning and Artificial Intelligence in healthcare.
Year(s) Of Engagement Activity 2017
URL https://theconversation.com/deepmind-can-we-ever-trust-a-machine-to-diagnose-cancer-88707
 
Description Brain'Art 2018 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Public/other audiences
Results and Impact Working with the Coma Science Group at the University of Liege, the PI created a visually rich and detailed representation of the brain network of the famous Tibetan buddhist, writer, philosopher and "world's happiest man" Matthieu Ricard. This scientific artwork, which is a large lenticular print contrasting the vivid brain network of Ricard during rest and deep meditation, is titled The Meditative Mind. It was selected for the Brain'Art 2018 competition and was displayed at the iconic train station at Liege, Belgium during October and November 2018.
Year(s) Of Engagement Activity 2018
URL https://braincouncil.be/en/brainart/brainart2018/24-srivas-chennu-university-of-kent-rajanikant.jpg/...
 
Description British Society of Rehabilitation Medicine Lecture 2017 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact The PI gave a lecture to rehabilitation experts and clinical practitioners at the 2017 annual meeting of the British Society of Rehabilitation Medicine, about the potential value of computational connectomics for assessing consciousness in their patient populations.
Year(s) Of Engagement Activity 2017
 
Description Cambridge Science Festival 2018 - Brilliant Brains 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Public/other audiences
Results and Impact The PI contributed to an event involving interaction, discussion and interpretation of the science of consciousness with the artist Sarah Harley. This ongoing collaboration contributed to the creation of an artistic work that was displayed at the Brilliant Brains exhibition at the Cambridge Science Festival 2018.
Year(s) Of Engagement Activity 2018
 
Description Holy Cross Hospital Conference 2018 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Professional Practitioners
Results and Impact At a training conference for clinical rehabilitation experts and allied health professionals, the PI disseminated knowledge of this research project and recent developments in healthcare technology for bedside brain network assessments of consciousness in patients with severe brain injury.
Year(s) Of Engagement Activity 2018
URL https://www.holycross.org.uk/courses-and-conferences.html
 
Description ITV News at 10 Interview 
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 The PI gave expert commentary about brain-computer interface research that could have utility for patients in locked-in states.
Year(s) Of Engagement Activity 2017
URL https://www.itv.com/news/2017-01-31/scientific-breakthrough-as-researchers-read-thoughts-of-complete...
 
Description Short film on the Mohawk of Consciousness 
Form Of Engagement Activity A broadcast e.g. TV/radio/film/podcast (other than news/press)
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Public/other audiences
Results and Impact The PI contributed to and was interviewed for this short film about the impact of my research for patients, their families and clinicians. This film was aired on KMTV in December 2018, and was also screened at events involving families, carers and clinical staff at neuro-rehabilitation centres in Kent.
Year(s) Of Engagement Activity 2018
URL https://vimeo.com/290663867/f504da3a5d