Control of a visual keyboard using an electrocorticographic brain-computer interface

Dean J. Krusienski, Jerry J. Shih

Research output: Contribution to journalArticle

25 Citations (Scopus)

Abstract

Objective. Brain-computer interfaces (BCIs) are devices that enable severely disabled people to communicate and interact with their environments using their brain waves. Most studies investigating BCI in humans have used scalp EEG as the source of electrical signals and focused on motor control of prostheses or computer cursors on a screen. The authors hypothesize that the use of brain signals obtained directly from the cortical surface will more effectively control a communication/spelling task compared to scalp EEG. Methods. A total of 6 patients with medically intractable epilepsy were tested for the ability to control a visual keyboard using electrocorticographic (ECOG) signals. ECOG data collected during a P300 visual task paradigm were preprocessed and used to train a linear classifier to subsequently predict the intended target letters. Results. The classifier was able to predict the intended target character at or near 100% accuracy using fewer than 15 stimulation sequences in 5 of the 6 people tested. ECOG data from electrodes outside the language cortex contributed to the classifier and enabled participants to write words on a visual keyboard. Conclusions. This is a novel finding because previous invasive BCI research in humans used signals exclusively from the motor cortex to control a computer cursor or prosthetic device. These results demonstrate that ECOG signals from electrodes both overlying and outside the language cortex can reliably control a visual keyboard to generate language output without voice or limb movements.

Original languageEnglish (US)
Pages (from-to)323-331
Number of pages9
JournalNeurorehabilitation and Neural Repair
Volume25
Issue number4
DOIs
StatePublished - May 2011

Fingerprint

Brain-Computer Interfaces
Language
Scalp
Electroencephalography
Electrodes
Brain Waves
Equipment and Supplies
Aptitude
Motor Cortex
Prostheses and Implants
Extremities
Communication
Brain
Research

Keywords

  • Brain-computer interface
  • Electrocorticography
  • Event-related potentials
  • P300 speller

ASJC Scopus subject areas

  • Clinical Neurology
  • Rehabilitation
  • Neurology

Cite this

Control of a visual keyboard using an electrocorticographic brain-computer interface. / Krusienski, Dean J.; Shih, Jerry J.

In: Neurorehabilitation and Neural Repair, Vol. 25, No. 4, 05.2011, p. 323-331.

Research output: Contribution to journalArticle

Krusienski, Dean J. ; Shih, Jerry J. / Control of a visual keyboard using an electrocorticographic brain-computer interface. In: Neurorehabilitation and Neural Repair. 2011 ; Vol. 25, No. 4. pp. 323-331.
@article{91026993f19744609abcb35258d747dd,
title = "Control of a visual keyboard using an electrocorticographic brain-computer interface",
abstract = "Objective. Brain-computer interfaces (BCIs) are devices that enable severely disabled people to communicate and interact with their environments using their brain waves. Most studies investigating BCI in humans have used scalp EEG as the source of electrical signals and focused on motor control of prostheses or computer cursors on a screen. The authors hypothesize that the use of brain signals obtained directly from the cortical surface will more effectively control a communication/spelling task compared to scalp EEG. Methods. A total of 6 patients with medically intractable epilepsy were tested for the ability to control a visual keyboard using electrocorticographic (ECOG) signals. ECOG data collected during a P300 visual task paradigm were preprocessed and used to train a linear classifier to subsequently predict the intended target letters. Results. The classifier was able to predict the intended target character at or near 100{\%} accuracy using fewer than 15 stimulation sequences in 5 of the 6 people tested. ECOG data from electrodes outside the language cortex contributed to the classifier and enabled participants to write words on a visual keyboard. Conclusions. This is a novel finding because previous invasive BCI research in humans used signals exclusively from the motor cortex to control a computer cursor or prosthetic device. These results demonstrate that ECOG signals from electrodes both overlying and outside the language cortex can reliably control a visual keyboard to generate language output without voice or limb movements.",
keywords = "Brain-computer interface, Electrocorticography, Event-related potentials, P300 speller",
author = "Krusienski, {Dean J.} and Shih, {Jerry J.}",
year = "2011",
month = "5",
doi = "10.1177/1545968310382425",
language = "English (US)",
volume = "25",
pages = "323--331",
journal = "Neurorehabilitation and Neural Repair",
issn = "1545-9683",
publisher = "SAGE Publications Inc.",
number = "4",

}

TY - JOUR

T1 - Control of a visual keyboard using an electrocorticographic brain-computer interface

AU - Krusienski, Dean J.

AU - Shih, Jerry J.

PY - 2011/5

Y1 - 2011/5

N2 - Objective. Brain-computer interfaces (BCIs) are devices that enable severely disabled people to communicate and interact with their environments using their brain waves. Most studies investigating BCI in humans have used scalp EEG as the source of electrical signals and focused on motor control of prostheses or computer cursors on a screen. The authors hypothesize that the use of brain signals obtained directly from the cortical surface will more effectively control a communication/spelling task compared to scalp EEG. Methods. A total of 6 patients with medically intractable epilepsy were tested for the ability to control a visual keyboard using electrocorticographic (ECOG) signals. ECOG data collected during a P300 visual task paradigm were preprocessed and used to train a linear classifier to subsequently predict the intended target letters. Results. The classifier was able to predict the intended target character at or near 100% accuracy using fewer than 15 stimulation sequences in 5 of the 6 people tested. ECOG data from electrodes outside the language cortex contributed to the classifier and enabled participants to write words on a visual keyboard. Conclusions. This is a novel finding because previous invasive BCI research in humans used signals exclusively from the motor cortex to control a computer cursor or prosthetic device. These results demonstrate that ECOG signals from electrodes both overlying and outside the language cortex can reliably control a visual keyboard to generate language output without voice or limb movements.

AB - Objective. Brain-computer interfaces (BCIs) are devices that enable severely disabled people to communicate and interact with their environments using their brain waves. Most studies investigating BCI in humans have used scalp EEG as the source of electrical signals and focused on motor control of prostheses or computer cursors on a screen. The authors hypothesize that the use of brain signals obtained directly from the cortical surface will more effectively control a communication/spelling task compared to scalp EEG. Methods. A total of 6 patients with medically intractable epilepsy were tested for the ability to control a visual keyboard using electrocorticographic (ECOG) signals. ECOG data collected during a P300 visual task paradigm were preprocessed and used to train a linear classifier to subsequently predict the intended target letters. Results. The classifier was able to predict the intended target character at or near 100% accuracy using fewer than 15 stimulation sequences in 5 of the 6 people tested. ECOG data from electrodes outside the language cortex contributed to the classifier and enabled participants to write words on a visual keyboard. Conclusions. This is a novel finding because previous invasive BCI research in humans used signals exclusively from the motor cortex to control a computer cursor or prosthetic device. These results demonstrate that ECOG signals from electrodes both overlying and outside the language cortex can reliably control a visual keyboard to generate language output without voice or limb movements.

KW - Brain-computer interface

KW - Electrocorticography

KW - Event-related potentials

KW - P300 speller

UR - http://www.scopus.com/inward/record.url?scp=79954593214&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79954593214&partnerID=8YFLogxK

U2 - 10.1177/1545968310382425

DO - 10.1177/1545968310382425

M3 - Article

C2 - 20921326

AN - SCOPUS:79954593214

VL - 25

SP - 323

EP - 331

JO - Neurorehabilitation and Neural Repair

JF - Neurorehabilitation and Neural Repair

SN - 1545-9683

IS - 4

ER -