Project

Back to overview

Error-adaptive decoding algorithms for stable and independent brain-computer interfaces

Applicant Milekovic Tomislav
Number 168103
Funding scheme Ambizione
Research institution Département des neurosciences fondamentales Faculté de Médecine Université de Genève
Institution of higher education University of Geneva - GE
Main discipline Other disciplines of Engineering Sciences
Start/End 01.04.2017 - 31.03.2020
Approved amount 577'447.00
Show all

All Disciplines (4)

Discipline
Other disciplines of Engineering Sciences
Information Technology
Biomedical Engineering
Electrical Engineering

Keywords (11)

Paralysis; Microelectrode array; Electrophysiology; Action potentials; Brain-Computer Interface; Adaptive algorithms; Motor cortex; Error-related neural responses; Tetraplegia; Locked-in syndrome; Local field potentials

Lay Summary (German)

Lead
Neue Ansätze für eine stabile und unabhängige Kontrolle von Hirn-Computer Schnittstellen basierend auf Fehler-bezogenen Hirnpotentialen
Lay summary

Paralyse beeinträchtigt die Lebensqualität der Patienten sehr, stellt eine hohe emotionale Last dar, und führt zu lebenslangen sozialen und finanziellen Kosten. Die Wiederherstellung von Bewegung und Unabhängigkeit bei Menschen mit schwerer Paralyse ist nach wie vor ein schwieriges klinisches Problem ohne echte Lösungen. In jüngster Zeit haben Demonstrationen von Hirn-Computer Schnittstellen vielen Patienten mit Paralysen Hoffnung gemacht einen Teil von Bewegung und Kommunikation wiederzuerlangen. Dies sind neuroprothetische Geräte die alleine durch Hirnaktivität eine Verbindung zwischen einer Person und einem Computer herstellen. Das Herz dieser Schnittstellen sind Algorithmen die kontinuierlich die Hirnaktivität dekodieren und in Computerbefehle umwandeln. Die Algorithmen passen sich automatisch den Veränderungen der Hirnaktivität an und halten so ihre autonome Leistungsfähigkeit auch wenn die Hirnsignale unstabil werden. Damit solche automatischen Anpassungen wirkungsvoll bleiben, müssen sie Fehler im Dekodieren korrekt erkennen. Solche Fehler können anhand von spezifischen Hirnsignalen erkannt werden die entstehen wenn das Verhalten des Computers nicht dem entspricht was man erwartet hat. Das Projekt wird Algorithmen entwickeln die diese vom Hirn generierten Fehler-Signale erkennt und ausnutzt um stabile und unabhängige Hirn-Computer Schnittstellen zu generieren. Dies wird uns einen Schritt weiter bringen in der Entwicklung von Hirn-Computer Geräten für den Einsatz im Alltag.

Direct link to Lay Summary Last update: 21.09.2016

Responsible applicant and co-applicants

Employees

Publications

Publication
Volitional control of single-electrode high gamma local field potentials by people with paralysis
Milekovic Tomislav, Bacher Daniel, Sarma Anish A., Simeral John D., Saab Jad, Pandarinath Chethan, Yvert Blaise, Sorice Brittany L., Blabe Christine, Oakley Erin M., Tringale Kathryn R., Eskandar Emad, Cash Sydney S., Shenoy Krishna V., Henderson Jaimie M., Hochberg Leigh R., Donoghue John P. (2019), Volitional control of single-electrode high gamma local field potentials by people with paralysis, in Journal of Neurophysiology, 121(4), 1428-1450.

Collaboration

Group / person Country
Types of collaboration
Wyss Center Switzerland (Europe)
- Research Infrastructure
Gregoire Courtine, EPFL Switzerland (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Research Infrastructure
François Lazeyras, Center for Biomedical Imaging (CIBM) Switzerland (Europe)
- Research Infrastructure
Niels Birbaumer, University of Tubingen Germany (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Research Infrastructure
Karin Diserens, University of Lausanne Switzerland (Europe)
- in-depth/constructive exchanges on approaches, methods or results
John Donoghue Switzerland (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
- Research Infrastructure
- Exchange of personnel
Ammar Kassouha, University of Geneva Switzerland (Europe)
- in-depth/constructive exchanges on approaches, methods or results
Bogdan Draganski, University of Lausanne Switzerland (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Research Infrastructure
Jocelyne Bloch Switzerland (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Research Infrastructure
Ruxandra Iancu-Ferfoglia, University of Geneva Switzerland (Europe)
- in-depth/constructive exchanges on approaches, methods or results

Scientific events

Active participation

Title Type of contribution Title of article or contribution Date Place Persons involved
Swiss Federation of Clinical Neuro-Societies congress 2019 Poster Brain-computer interfaces based on cortical source activity during attempted movements reconstructed from high-density EEG in patients with amyotrophic lateral sclerosis 23.10.2019 Lausanne, Switzerland Pokorny Christoph;
BBL/CIBM/FCBG MRI research day 2019 Talk given at a conference Communication brain-computer interfaces for people with locked-in syndrome 27.05.2019 Geneva, Switzerland Milekovic Tomislav; Pokorny Christoph;
European Stroke Organisation Conference 2019 Talk given at a conference Brain-controlled assistive devices for people with brainstem stroke 24.05.2019 Milan, Italy Milekovic Tomislav;
Wings for Life Scientific conference 2019 Talk given at a conference Brain-controlled spinal cord stimulation to alleviate gait deficits in people with paraplegia 08.05.2019 Salzburg, Austria Milekovic Tomislav;
Swiss Society for Neuroscience Annual Meeting 2019 Poster Motor-cortical responses during attempted movements in patients with amyotrophic lateral sclerosis reconstructed from high-density EEG 01.02.2019 Geneva, Switzerland Pokorny Christoph;
1st Swiss Early-career Researchers Symposium Talk given at a conference Reconstructed activity of cortical sources during attempted movements in patients with amyotrophic lateral sclerosis 31.01.2019 Geneva, Switzerland Pokorny Christoph;
Alpine Brain Imaging Meeting 2019 Poster Asynchronous decoding of attempted movements for brain-computer interfaces using source reconstructed cortical potentials 06.01.2019 Champery, Switzerland Pokorny Christoph;
3rd Annual Wyss Investigators Symposium Talk given at a conference Enhancing non-invasive brain-computer interfaces for communication in people with locked-in syndrome by combining high-density EEG and self-calibrating algorithms 06.12.2018 Geneva, Switzerland Milekovic Tomislav;
Neuroscience 2018 American Society for Neuroscience annual meeting Poster Utility and versatility of an asynchronous switch neural decoders based on regularized Gaussian Mixture Model for neuroprosthetic applications 03.11.2018 San Diego, CA, United States of America Milekovic Tomislav;
MoBI Conference 2018 Talk given at a conference Using EEG source imaging to study gait-related cortical dynamics 11.07.2018 Berlin, Germany Seeber Martin;
Bernstein Center Fribourg Seminar Individual talk Neuroprostheses based on intracortical recordings of neural activity for restoration of movement and communication of people with paralysis 24.04.2018 Freiburg, Germany Milekovic Tomislav;
6th International Conference on Brain-Computer Interface (BCI) 2018 Talk given at a conference Brain-computer interfaces based on intracortical recordings of neural activity for restoration of movement and communication of people with paralysis 15.01.2018 Gohan, Korean Republic (South Korea) Milekovic Tomislav;
Swiss Paraplegic Centre Seminar Individual talk Neuroprostheses based on intracortical recordings of neural activity for restoration of movement and communication of people with paralysis 21.12.2017 Notwill, Switzerland Milekovic Tomislav;
European Congress of NeuroRehabilitation 2017 Poster Neuroprostheses based on intracortical recordings of neural activity for restoration of movement and communication of people with paralysis 25.10.2017 Lausanne, Switzerland Milekovic Tomislav;
Final ANDREA Workshop Talk given at a conference EEG brain imaging during gait 30.08.2017 Bern, Switzerland Seeber Martin;
BaCI Conference 2017 Poster From repetitive to brisk movements: EEG source features for Brain-Computer Interfaces 29.08.2017 Bern, Switzerland Milekovic Tomislav; Seeber Martin;
BaCI Conference 2017 Talk given at a conference Neuroprostheses based on intracortical recordings of neural activity for restoration of movement and communication of people with paralysis 29.08.2017 Bern, Switzerland Milekovic Tomislav;
Kavli Foundation seminar organized at the Wyss Center for Bio and Neuroengineering Individual talk Evolution of large-scale cortical networks in people with neurodegenerative disorders 22.08.2017 Geneva, Switzerland Milekovic Tomislav;
Fondation Campus Biotech Geneva Project Presentation Seminar Individual talk Combining source reconstruction and error-adaptive decoding algorithms for EEG-based BCIs 03.07.2017 Geneva, Switzerland Milekovic Tomislav;
Innovation center for Assistive Technologies Seminar Individual talk Neuroprosthetic technologies to restore movement and communication of people with paralysis 06.04.2017 Neuchatel, Switzerland Milekovic Tomislav;


Communication with the public

Communication Title Media Place Year
Media relations: print media, online media Dizajnirao neuroprotezu koja je majmunu vratila kontrolu nad paraliziranom nogom i dao nadu osobama Jutarnji List International 2017
Media relations: print media, online media SADA RADIM NA NEUROPROTEZI KOJA ĆE UBRZO PARALIZIRANIMA VRATITI POKRETE Jutarnji List International 2017

Awards

Title Year
International BCI award finalist 2019
International BCI award finalist 2017
PIs of Tomorrow competition finalist 2017

Abstract

Recent demonstrations of brain-computer interfaces (BCIs), neuroprosthetic devices that create a link between a person and a computer based on person’s brain activity, have brought hope to millions of people with paralysis for their potential to restore movement and communication. This project aims to develop approaches for independent and stable control of BCIs, thereby removing a major obstacle to translating BCIs from clinical demonstrations into real-world assistive devices.BCIs based on recordings from intracortically implanted microelectrode arrays provide the most accurate control of computer systems, but require surgical implantation. Electroencephalography (EEG) based BCIs utilize whole brain activity and do not require surgeries, but have lower performance than “intracortical” BCIs. Nonetheless, combining newly developed high-density EEG systems with distributed EEG source analysis may boost performance. Advancing both of these systems to real-world applications will rely on the ability to generate accurate and stable BCI control from the continuously recorded neural signals. Nevertheless, a range of factors can lead to unstable neural signals, which then lead to progressive deterioration of BCI control. Overcoming this problem by adaptive decoding algorithms is the major goal of this project.A BCI decoding algorithm maps continuously recorded neural signals to interface commands. Adaptive decoding algorithms can automatically adjust to neural signal changes, thus maintaining BCI performance despite unstable neural signals and without the interference of technicians. However, effective adaptation requires accurate detection of decoding errors. Decoding errors can be detected from error-related neural responses, evoked by a mismatch between an expected and observed command outcome. Recent studies have used error-related EEG responses, mainly recorded by one EEG electrode, to adapt BCI decoding algorithms. Distributed EEG source analysis may be used to more precisely identify error-related EEG responses and more accurately detect decoding errors. Moreover, error-related neural responses have not yet been identified in intracortical recordings from the motor cortex, signals predominantly used to achieve intracortical BCIs. To achieve long-term stable BCI performance, we will (i) identify error-related neural responses in high-density EEG and intracortical neural recordings and (ii) develop, implement and evaluate a unique error-adaptive decoding algorithm based on real-time recognition of error-related neural responses. To this end, we will conduct a study with healthy volunteers and people with tetraplegia. First, all of the study participants will interact with a computer task designed to elicit error-related neural responses (error task) in order to identify error neural sources using distributed EEG source analysis. Second, the participants will use an EEG-based communication BCI with a newly developed error-adaptive algorithm for over three months without technical intervention to demonstrate its stability and independence. Decoding errors will be detected in real-time using the activity of identified error neural sources. Third, participants with tetraplegia will be intracortically implanted with microelectrode arrays in the motor cortex. They will again perform the error task, this time to identify error-related intracortical neural responses. Fourth, they will use the same communication BCI for over three months, this time with the error-adaptive algorithm based on intracortically recorded neural signals. In addition to developing independent and stable BCIs, this project will make the first direct and comprehensive comparison between EEG and intracortically-based BCIs, thereby informing the future BCI users of performance versus invasiveness tradeoffs.
-