Direct neural interface system and method of calibrating it

Patent metadata

Field	Value
Publication number	US-9480583-B2
Application number	US-201013698166-A
Country	US
Kind code	B2
Filing date	May 17, 2010
Priority date	May 17, 2010
Publication date	Nov 1, 2016
Grant date	Nov 1, 2016

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.

First claim

The invention claimed is:

1. A direct neural interface system comprising:
signal acquisition mechanism configured for acquiring electrophysiological signals representative of a neuronal activity of a subject's brain (B);
preprocessor (PPM) for conditioning, digitizing and preprocessing said electrophysiological signals;
processor configured (PM) for processing the digitized and preprocessed electrophysiological signals and for generating command signals as a function thereof; and
output for outputting said command signals configured for driving an external device;
wherein said preprocessor and said processor are configured for:
generating a N-way data tensor representing the electrophysiological signals acquired over an observation time window, N being greater or equal to two; and
generating said command signals configured for driving the external device, wherein said command signals correspond to said observation time window by applying a multi-way regression model over said data tensor, and
wherein said processor is configured for generating an output signal corresponding to said observation time window by applying a multi-way partial least square—or NPLS—regression model to said data tensor, which allows better utilization of available information and avoids the need for human intervention for associating electrophysiological activity to intended or motor action.

2. A direct neural interface system according to claim 1 , wherein said processor is configured for representing the electrophysiological signals acquired over an observation time window in the form of a three-way data tensor, said three ways corresponding to time, frequency and space.

3. A direct neural interface system according to claim 1 , wherein said signal acquisition mechanism is configured for acquiring ECoG signals.

4. A method of calibrating a direct neural interface system according to claim 1 , comprising the steps of:
a. acquiring said electrophysiological signals over a plurality of observation time windows and generating a N+1-way observation tensor representing said acquired electrophysiological signals, N being greater or equal to two;
b. acquiring data indicative of a voluntary action performed by said subject during each of said observation time windows, and organizing them in an output tensor; and
c. determining a multi-way regression function of said output tensor on said observation tensor, wherein said steps of acquiring electrophysiological signals, acquiring data indicative of a voluntary action, and determining a multi-way regression function are performed by said direct neural interface system.

5. A method according to claim 4 , wherein said step c. includes determining a multilinear regression function of said output tensor on said observation tensor.

6. A method according to claim 4 , wherein said step c.includes performing multilinear decomposition of said observation tensor on a “score” vector, having a dimension equal to the number of said observation time windows, and N “weight” vectors.

7. A method according to claim 6 , wherein said “weight” vectors are chosen such as to maximize a covariance between said “score” vector and said output tensor.

8. A method according to claim 7 , wherein said step c. includes performing PARAFAC decomposition of a covariance tensor, representing a covariance of said observation tensor and said output tensor, yielding said “weight” vectors.

9. A method according to claim 8 , wherein said step c. includes:
c1. subdividing the observation tensor and the output tensor in a plurality of smaller tensors, each corresponding to a subset of said observation time windows;
c2. taking a set of predetermined or random vectors as a first estimation of the “weight” vectors resulting from decomposition of said covariance tensor;
c3. successively decomposing the subdivisions of said observation tensor, thus obtaining respective partial “weight” vectors, each decomposition being initialized by a current estimation of the “weight” vectors; and using said set of partial “weight” vectors for improving said estimation before performing a subsequent decomposition.

10. A method according to claim 4 , wherein said observation tensor is a three-way data tensor, said three ways corresponding to time, frequency and space.

11. A method of interfacing a subject's brain to an external device by a direct neural interface system according to claim 1 , said method comprising the steps of:
acquiring, conditioning, digitizing and preprocessing electrophysiological signals representative of a neuronal activity of said subject's brain over at least one instance of said observation time window; and
generating said command signals for said external device by processing said digitized and preprocessed electrophysiological signals;
wherein said step of generating command signals comprises:
generating each said N-way data tensor representing the electrophysiological signals acquired over each instance of said observation time window, N being greater or equal to three; and
generating an output signal corresponding to each instance of said observation time window by performing multi-way partial least squares—or NPLS—regression over each said data tensor,
wherein said steps of acquiring, conditioning, digitizing, and preprocessing electrophysiological signals and generating said command signals are performed by said direct neural interface system.

12. A method according to claim 11 , comprising generating each said N-way data tensor representing the electrophysiological signals acquired over each instance of said observation time window in the form of a three-way data tensor, said three ways corresponding to time, frequency and space.

13. A method according to claim 11 , wherein said electrophysiological signals are ECoG signals.

14. A method according to claim 11 , comprising a calibration step comprising the steps of:
a. acquiring said electrophysiological signals over a plurality of observation time windows and generating a N+1-way observation tensor representing the acquired electrophysiological signals, N being greater or equal to two;
b. acquiring data indicative of a voluntary action performed by said subject during each of said observation time windows, and organizing them in an output tensor; and
c. determining a multi-way regression function of said output tensor on said observation tensor.

15. A method according to claim 11 , wherein the generation of said command signals is self-paced.

Assignees

• Aksenova Tetiana
• Yelisyeyev Andriy
• Commissariat Energie Atomique

Inventors

• Aksenova Tetiana
• Yelisyeyev Andriy

Classifications

• A61B5/0476

  Human Necessities · mapped topic

• A61B5/726

  using Wavelet transforms · CPC title

• A61B5/04012

  Human Necessities · mapped topic

• A61F2/72 Primary

  Bioelectric control, e.g. myoelectric · CPC title

• A61F4/00

  Methods or devices enabling patients or disabled persons to operate an apparatus or a device not forming part of the body · CPC title