Abstract: The present invention relates to a method for calibrating on-line a direct neural interface implementing a REW-NPLS regression between an output calibration tensor and an input calibration tensor. The REW-NPLS regression comprises a PARAFAC iterative decomposition of the cross covariance tensor between the input calibration tensor and the output calibration tensor, each PARAFAC iteration comprising a sequence of M elementary steps (2401, 2401, . . . 240M) of minimisation of a metric according to the alternating least squares method, each elementary minimisation step relating to a projector and considering the others as constant, said metric comprising a penalisation term that is a function of the norm of this projector, the elements of this projector not being subjected to a penalisation during a PARAFAC iteration f not being penalisable during following PARAFAC iterations. Said calibration method makes it possible to obtain a predictive model of which the non-zero coefficients are sparse blockwise.

Abstract: The invention relates to a method for controlling the movement, through a space, of an effector via a brain-computer interface, comprising: a) acquiring electrophysiological signals produced in the cortex of an individual, at a measurement time; b) processing the electrophysiological signals, to form input data; c) processing the input data, by means of a predictive model implemented by the brain-computer interface, to define an effector movement (xps) at the measurement time; d) determining a corrected effector movement (xcs+1?xcs), based on the movement (xps) defined by the predictive model in step c); e) moving the effector through the space, by means of the brain-computer interface, based on the corrected movement resulting from step d); f) reiterating steps (a) to (e), the method comprising, in each step d), estimating the target position , at the measurement time. FIG. 2.

Abstract: The present invention relates to an adaptive training method of a brain computer interface. The ECoG signals expressing the neural command of the subject are preprocessed to provide at each observation instant an observation data tensor to a predictive model that deduces therefrom a command data tensor making it possible to control a set of effectors. A satisfaction/error mental state decoder predicts at each epoch a satisfaction or error state from the observation data tensor. The mental state predicted at a given instant is used by an automatic data labelling module to generate on the fly new training data from the pair formed by the observation data tensor and the command data tensor at the preceding instant. The parameters of the predictive model are subsequently updated by minimising a cost function on the training data thus generated.

Abstract: The present invention relates to a method for calibrating on-line a direct neural interface implementing a REW-NPLS regression between an output calibration tensor and an input calibration tensor. The REW-NPLS regression comprises a PARAFAC iterative decomposition of the cross covariance tensor between the input calibration tensor and the output calibration tensor, each PARAFAC iteration comprising a sequence of M elementary steps (2401, 2401, . . . 240M) of minimisation of a metric according to the alternating least squares method, each elementary minimisation step relating to a projector and considering the others as constant, said metric comprising a penalisation term that is a function of the norm of this projector, the elements of this projector not being subjected to a penalisation during a PARAFAC iteration f not being penalisable during following PARAFAC iterations. Said calibration method makes it possible to obtain a predictive model of which the non-zero coefficients are sparse blockwise.

Abstract: This invention relates to a method of calibrating a direct neural interface with continuous coding. The observation variable is modelled by an HMM model and the control variable is estimated by means of a Markov mixture of experts, each expert being associated with a state of the model. During each calibration phase, the predictive model of each of the experts is trained on a sub-sequence of observation instants corresponding to the state with which it is associated, using an REW-NPLS (Recursive Exponentially Weighted N-way Partial Least Squares) regression model. A second predictive model giving the probability of occupancy of each state of the HMM model is also trained during each calibration phase using an REW-NPLS regression method. This second predictive model is used to calculate Markov mixture coefficients during a later operational prediction phase.

Abstract: A method of continuous decoding of motion for a direct neural interface. The method of decoding estimates a motion variable from an observation variable obtained by a time-frequency transformation of the neural signals. The observation variable is modelled using a HMM model whose hidden states include at least an active state and an idle state. The motion variable is estimated using a Markov mixture of experts where each expert is associated with a state of the model. For a sequence of observation vectors, the probability that the model is in a given state is estimated, and from this a weighting coefficient is deduced for the prediction generated by the expert associated with this state. The motion variable is then estimated by combination of the estimates of the different experts with these weighting coefficients.

Abstract: A direct neural interface system comprises: a signal acquisition subsystem for acquiring electrophysiological signals representative of neuronal activity of a subject's brain; and a processing unit for representing electrophysiological signals acquired over an observation time window in the form of a N-way data tensor, N being greater than or equal to two, and generating command signals for a machine by applying a regression model over the data tensor; wherein the processing unit is configured or programmed for generating command signals for a machine by applying Generalized Linear regression, with a nonlinear link function, over the data tensor. A method of interfacing a subject's brain to a machine by using such a direct neural interface system is provided.

Abstract: A method of continuous decoding of motion for a direct neural interface. The method of decoding estimates a motion variable from an observation variable obtained by a time-frequency transformation of the neural signals. The observation variable is modelled using a HMM model whose hidden states include at least an active state and an idle state. The motion variable is estimated using a Markov mixture of experts where each expert is associated with a state of the model. For a sequence of observation vectors, the probability that the model is in a given state is estimated, and from this a weighting coefficient is deduced for the prediction generated by the expert associated with this state. The motion variable is then estimated by combination of the estimates of the different experts with these weighting coefficients.

Abstract: A direct neural interface system comprised of electrodes for acquiring electrophysiological signals representative of a neuronal activity of a subject's brain; a pre-processor for conditioning, digitizing and preprocessing the electrophysiological signals; a processor for processing the digitized and preprocessed electrophysiological signals and generating command signals; and an output for outputting said command signals; wherein the processor is adapted for: representing the electrophysiological signals acquired over an observation time window in the form of a N-way data tensor, N being greater or equal to three; and generating command signals corresponding to the observation time window by applying a multi-way regression model over the data tensor. A method of calibrating the direct neural interface system.

Abstract: A direct neural interface system comprises: a signal acquisition subsystem for acquiring electrophysiological signals representative of neuronal activity of a subject's brain; and a processing unit for representing electrophysiological signals acquired over an observation time window in the form of a N-way data tensor, N being greater than or equal to two, and generating command signals for a machine by applying a regression model over the data tensor; wherein the processing unit is configured or programmed for generating command signals for a machine by applying Generalized Linear regression, with a nonlinear link function, over the data tensor. A method of interfacing a subject's brain to a machine by using such a direct neural interface system is provided.

Abstract: Method for locating a brain activity, including the following steps: a) applying to a subject a first series of sensory stimuli and acquiring, by a group of sensors, respective first series of signals representative of a brain activity associated with a first task effected or imagined by the subject in response to the sensory stimuli of the first series, each sensor being sensitive to the activity of a respective region of the brain of the subject; b) applying to the subject a second series of sensory stimuli and acquiring, by the group of sensors, respective second series of signals representative of a brain activity associated with a second task, different from the first task, effected or imagined by the subject in response to the sensory stimuli of the second series; and c) constructing, for each sensor, a multidimensional variable representative of the corresponding first and second series of signals, and determining a coefficient of correlation between the multidimensional variable and an observation vec

Abstract: Method for determining at least one wavelet coefficient Ws(?) of a wavelet transform of a signal in which the mother wavelet of the transform has a support subdivided into J?1 intervals bound by (J+1) extremity points, and is defined by a polynomial of a maximum level N?1 on each interval. The method includes calculating all or some of the primitives of the signal of order k between 2 and N+1, at least at (J+1) points corresponding to extremity points of the intervals of the wavelet support dilated by a factor of s and translated by a time ?; calculating the convolution of said or each primitive sampled in this way with a respective succession of (J+1) coefficients Cik(s), dependent upon said wavelet; and determining the wavelet coefficient by calculating a linear combination of convolutions. Steps a) to c) are implemented by a processor configured or programmed in an appropriate manner.

Abstract: Method for locating a brain activity, comprising the steps consisting in: a) acquiring data indicative of sensory stimuli addressed to, or deliberate actions performed or imagined by, a subject; b) by means of a plurality of sensors, acquiring signals representative of an activity, associated with said stimuli or deliberate actions, of respective regions of the brain of said subject; and c) for each said sensor, quantifying a correlation that exists between said data indicative of sensory stimuli or of deliberate actions and the signals acquired; characterized in that, for each said sensor, said correlation is established between a scalar variable indicative of a said sensory stimulus or of a said deliberate action and a multidimensional variable representative of signals that are associated therewith. Application of this method to the determination of optimum locations for brain activity sensors for direct neural control.

Abstract: The present invention relates to a method for filtering the signal of neuronal activity during a high frequency deep brain stimulation (DBS) to remove the stimulus artefact in the observed signal, comprising the step of approximating the observed signal trajectories in phase space the observed signal being considered as a sum of the stimulation artifacts induced by the signal of stimulation, wherein the signal of stimulation is assumed to be a solution of an ordinary differential equation including a self-oscillating system with stable limit cycle; slicing the observed signal and its derivative into segments, each segment corresponding to a period of stimulation; collecting N selected periods of stimulation to a training set; estimating the limit cycle of the self-oscillating system; synchronizing each artefact of the observed signal with the estimated limit cycle; subtracting the estimated limit cycle from each artefact in phase space according to the synchronization; collecting all segments in order to obta

Abstract: A direct neural interface system comprising: signal acquisition means (2-15) for acquiring electrophysiological signals representative of a neuronal activity of a subject's brain (B); preprocessing means (PPM) for conditioning, digitizing and preprocessing said electrophysiological signals; processing means (PM) for processing the digitized and preprocessed electrophysiological signals and for generating command signals as a function thereof; and output means for outputting said command signals; characterized in that said processing means are adapted for: representing the electrophysiological signals acquired over an observation time window in the form of a N-way data tensor, N being greater or equal to three; and generating command signals corresponding to said observation time window by applying a multi-way regression model over said data tensor. A method of calibrating said direct neural interface system.

Abstract: The present invention relates to a method for filtering the signal of neuronal activity during a high frequency deep brain stimulation (DBS) to remove the stimulus artefact in the observed signal, comprising the step of approximating the observed signal trajectories in phase space the observed signal being considered as a sum of the stimulation artefacts induced by the signal of stimulation, wherein the signal of stimulation is assumed to be a solution of an ordinary differential equation including a self-oscillating system with stable limit cycle; slicing the observed signal and its derivative into segments, each segment corresponding to a period of stimulation; collecting N selected periods of stimulation to a training set; estimating the limit cycle of the self-oscillating system; synchronizing each artefact of the observed signal with the estimated limit cycle; subtracting the estimated limit cycle from each artefact in phase space according to the synchronization; collecting all segments in order to obta