Aldemar Torres Valderrama
Position:
Senior Post-docGrants
BrainGain 2008-2010
VENI 2010-2013 (personal grant)
Duration project
August 1st 2008 - August 1st 2012
Project 1
Epidural vs subdural brain signal recordings in humans
Content
Invasive brain computer interfaces use the brain neural activity recorded intracranially with multi-electrode arrays to control artificial devices. These methods have strong advantages compared to scalp recordings in terms of band width (0-200Hz vs 0-40Hz), spatial resolution (millimeters vs centimeters) and signal quality. In addition, invasive recordings are less vulnerable to artifacts. Currently most invasive brain computer interfacing (BCI) studies are performed subdurally in patients with epilepsy before they undergo surgery. Subdural recordings have the disadvantage of reducing the long term stability of the implants and enhance the risk of infection. Epidural recordings for BCI purposes might offer a viable alternative to subdural implants, offering a compromise between signal quality and invasiveness. In this project we study to what extent epidurally recorded signals are sufficiently clear, undistorted and artifact free to be used for control/feedback brain computer interfacing applications. In particular, we measure the transfer function of the human dura.
Cooperation within UMC Utrecht
Nick Ramsey
Geertjan Huiskamp
Piere Gosselaar
Cooperation outside UMC Utrecht
Robert Oostenveld
Donders Institute for Brain, Cognition and Behaviour
P.O.Box 9101
NL-6500 HB Nijmegen
The Netherlands
Project 2
Sensor fusion by coherence features for brain computer interfacing
Content
The aim of the proposed research is to develop a novel invasive brain computer interface paradigm by studying synchronization among groups of neurons in the brain cortex during motor imagery. Invasive BCIs (see paragraph 1c) use signals recorded from the cerebral cortex by multi-electrode grids. Groups of neurons in the cortex exchange information by synchronizing their firing rates. Such coordinated activity (as detected in recorded signals) can (potentially) be used for BCI purposes. This project focuses on detecting, studying and utilizing synchronization of invasively recorded brain signals as control features for brain computer interfacing. The first key objective of this project is to detect synchronization patterns in invasively recorded signals associated with motor imagery by using statistical linear and non-linear signal feature extraction methods. The second key objective is to understand the origin of mesoscopic neural cooperativity as detected in the first key objective, from the perspective of neuron populations, by using theoretical methods of brain dynamics. The third key objective is to implement algorithms for BCI control based on the milestones established during the completion of the first and second key objectives, and to tests the algorithms online in invasive BCI experiments.
Cooperation within UMC Utrecht
Nick Ramsey
Cooperation outside UMC Utrecht
Robert Oostenveld
Donders Institute for Brain, Cognition and Behaviour
P.O.Box 9101
NL-6500 HB Nijmegen
The Netherlands
Project 3
Cortico-cortical coherence during imaginary and overt hand movement
Content
The objective of this research project is to detect synchronization patterns in invasively recorded signals during hand motor imagery and overt hand movement. Synchronization features originate themselves in the dynamics of functionally coupled brain areas and their chronological order of activation. In electroencephalography synchronization features have been interpreted as evidence for anatomical connections, functional coupling, information exchange, functional coordination, and temporal coordination between the cortical structures underlying these areas. Cortical and subcortical regions involved in motor control are activated when an individual moves or mentally simulates a motor action. Typically a decrease (with respect to idle states e.g., during states of immobility) is observed in the amplitude of signals recorded from sensory or motor areas in the frequency band of 8-12 Hz. Likewise, signals recorded invasively from the motor cortex exhibit changes in their synchronization properties across different channels. In this research project we use statistical linear and non-linear signal feature extraction methods in order to detect, localize and quantify such synchronization properties. We focus on electrocorticographic brain signals recorded from an epileptic patient during self-paced overt hand movement and motor imagery with feedback.
Cooperation within UMC Utrecht
Nick Ramsey
Cooperation outside UMC Utrecht
Alistair Fardy
Andreas Daffertshoffer
Faculty of Human movement science,
Vrije Universiteit, Amsterdam
Project 4
Correlation features of timed neural firing patterns
Content
During the execution of rhythmic movements, the individual movements are always subject to variation. Sometimes this variation may be intended, as in musical phrases that differ for expressive effect, but mostly it is involuntary. For rhythmic movements, short-term correlation patterns, e.g., negative lag-one serial correlations, are commonly found for a variety of movements. The two-level model as proposed by Wing and Kristofferson (Wing & Kristofferson, 1973) is often used to account for the serial correlations between successive time intervals found in behavioral data. The model consists of a timer or clock, defining the duration of the interval between two neural commands, and motor delays, which represent the time it takes for a neural command to reach the end-effector, i.e. the lag at which the tap occurs. The model does not require any assumptions regarding the origin of the clock, only that its timing intervals are independent. Although the model explains behavioral data well (Vorberg & Wing, 1996, Vardy et al., 2008), the temporal properties of the timer have not been verified. To accomplish this one may look for timed neural firing patterns with congruent (or different) correlation features one can discriminate between sources of (pure) variation and eventual error-correction mechanisms. To analyze neural activity in high temporal detail, one has a limited choice of methods. With this constraint in mind, EEG and MEG are commonly used, non-invasive techniques. A major drawback of these techniques, however, is the poor signal-to-noise ratio (in single event analysis), which renders clear-cut results challenging. Electrocorticography (ECoG) may help overcoming this difficulty as signals experience only few confounders (e.g., muscle artifacts are absent and volume conduction is limited). This invasive technique is used in patients with severe forms of epilepsy to locate the focus of epileptic activity in the brain. ECoG electrodes are placed directly on the surface of the brain. We believe that this technique has a sufficient temporal resolution and signal-to-noise ration as to give adequate insight into the timing of rhythmic movements on a cortical level, and elucidate the properties of the postulated neural timers.
Cooperation within UMC Utrecht
Nick Ramsey
Cooperation outside UMC Utrecht
Alistair Fardy
Andreas Daffertshoffer
Faculty of Human movement science,
Vrije Universiteit, Amsterdam
