Brain-Computer Interface

Primary objective of the BCI research program is to develop and implement an implantable BCI system for expression of intent in people who are locked in due to spinal cord lesions, brainstem stroke or severe motoneuron disease. We believe that it is possbile to record signals from very specific cortical sites, presurgically localized with custom-made high-accuracy fMRI techniques, and to decode signals therein generated deliberately by the user well enough to operate software designed for handicapped people (Assistive Technologies). This first step is expected to be a breakthrough in BCI, in terms of a) providing locked-in people with a means of communicating independent from a care giver, and b) have this tool avalilable at all times. This requires a multidisciplinary effort, ranging from rehabilitation medicine and cognitive neuroscience to Magnetic Resonance physics, state-of-the-art signal processing, hard- and software engineering.

The longer-term goals are to, once an implantable system has proven to function, increase the number of control channels and the richness of interfacing between a user and his/her environment. Ultimately people with full or even partial paralysis will be able (excuse my faith) to control their own limbs, by means of biocompatible electronics ('rewiring').The endeavours of the RIBS BCI group cover a specific corner of the BCI field. We specialize in identifying brain regions for electrode placement on the basis of the latest insights in human brain function in the Cognitve Neuroscience community, in teaching users how to regulate their brain activity in those regions, and in methods for decoding the neuronal signals acquired from those target regions. For hardware (electrodes, implantable micro-electronics) and software (Assistive Technology) we collaborate with other research groups and companies. This strategy enhances the feasibility of a commercially viable BCI implant solution.

The research is funded by multiple sources, and consists of several projects. These are detailed below.

Brain Computer Interface (VICI)

Title: Human CNS neuroprosthetics: substituting lost brain function.

Funding: STW (project 07685)
People: Mariska van Steensel (postdoc), Erik Aarnoutse (postdoc), Dora Hermes (phD student).

Abstract: People who are completely paralyzed, for instance due to a neuromuscular disease or spinal cord lesion, are often completely dependent on others. The goal of this research project is to develop neural prostheses (called Brain-Computer Interfaces or BCI's) that can be implanted in or on the brain of these patients, and enable them to move a cursor around in computer programs by merely thinking about making a movement. The neural prostheses read electrical signals directly from the brain, and convert it into a movement of the cursor, making it possible to communicate and to controle devices such as a television or a wheelchair. We will investigate which parts of the brain give the best results, and localize them with functional MRI scans. A second goal is to prevent loss of brainfunction in patients who need surgical removal of brain tumours, by stimulating other parts of the brain to take over that function.


Brain Computer Interface (Smartmix)

Title: The Power of Intracranial EEG for BCI.

Funding: Smartmix, Program BrainGain, Project 3
People: Aldemar Torres (postdoc), Jeroen Siero (PhD student), Martin Bleiochner (PhD student).

Abstract: Current Technology enables measurements of brain acitivity at many levels, ranging from single-cell recordings to whole-brain functional neuroimaging (eg fMRI). It is not clear which levels provide the best signals for successful BCI systems. There are tradeoffs in signal properties and invasiveness that are highly relevant to human BCI applications, and which need to be fully understood before considering brain surgery for BCI. Electrodes positioned in or on the cortical surface are widely expected to yield very localized brain signals with a high information content, both of which dramatically exceed those of scalp EEG systems. The exciting promise of intracranial electrodes is that they can be used for control of multiple devices. In this work package the various levels of recordings will be assessed and compared. Both animals and patients with implanted electrode grids will be studied, in a close collaboration between basic and clinical research groups. At the UMC Utrecht (prof. Ramsey) human studies focus on the signal properties of grid-electodes. These grids are implanted for accurate identification of epileptogenic brain tissue before surgical removal. Thus, electrodes are implanted for a medical reason, providing a excellent opportunity to study to potential use of intracranial electrodes for BCI. A major challenge is faced with regard to placing electrodes on specific brain areas without requiring major, open, brain surgery ro cover the brain with electrode grids. When target areas are identified as is planned in various other work packages, one would like to map that area on the surface of the head and accurately implant an electrode through a small hole in the skull. The human research also focuses on presurgical localization of brain regions that lend themselves to BCI applications (eg motor, language areas), in anticipation of intracranial BCI systems. The current most accurate method for localization is functional MRI, but this is not yet accurate enough to guide surgical electrode implantation. We aim to improve accuracy by modifying existing fMRI technology, using high-field (3-7 Tesla) MR systems.


Brain Computer Interface (UMC Utrecht)

Title: Use of Network Communication in Cognitive Systems for BCI.

Funding: Board of Directors, UMC Utrecht
People: Formerly Martin Bleichner (PhD student), position now open

Abstract: One of the unexplored issues in invasive BCI is the communicaton between different brain regions during working memory. Working memory is our niche in the field of BCI and we already have very promising results in grid patients. Currently centers only utilize signals from single elecrodes, and even there much of the signal processing possibilities are not yet fully explored, leaving room for better results. Already we have repeatedly achieved better results than what researchers get with scalp electrodes, supporting justification of further developing invasive BCI solutions. However, as we know from fMRI research much brain activity information can be found not only in acitivity in specific areas, but also in how brain areas within their functional networks communicate with each other. A recent development in MEG research is in mathematical representations of cross-regional communication. Notably, and originally developed in primate research, neural signals assume phase coherence when specific brain functions are invoked (eg a motor act, a working memory instance or a language processing instance). I postulate that phase coherence is a signature of very specific processes, and that subtle differences in brain functions are reflected by different signature of phase coherence between multiple regions of a network. The essential motive for this is that a good BCI solution requires multiple channels of control (eg move a cursor left, right up and down which constitutes 4 channels). Currently, groups obtain multiple channels by tapping different types of regions, such as left and right hand region, left and right foot, with limited success. This has a limited future because moving the cursor around then requires a considerable proficiency of the patient in imagining combined mental acts. Many patients are not very good at this. A better solution would be to obtain multiple channels that stem from natural distinctions between intentions. For instance, the cursor should move left when the patient thinks the word 'left', and right when he thinks 'right'. We are still far away from being able to achieve that, but this basic concept can be researched. In this project we will investigate: 1) optimal measurement and identification of phase coherence between two or more regions (i.e. mathematical computation and modeling), 2) classification of different coherence signatures for slightly different types of brain processes (eg thinking of a verb versus thinking of a noun), 3) comparison of coherence information to isolated activity (synchronization like gamma) information of the same regions, 4) the ability to use phase coherence from multi-channel BCI. We will perform the research in grid patients. The PhD student will focus on signals from multiple electrodes, obtained during performance of carefully designed cognitive paradigms, and will assess the benefits of using coherence measures.