Brain-computer interface: data collection and EEG waves

Brain-Computer Interfaces

Établi : June 29, 2018

Brain-Computer Interface (BCI) is a system that measures central nervous system (CNS) activity and converts it into artificial output that replaces, restores, enhances, supplements, or improves the natural CNS output and thereby changes the ongoing interactions between the CNS and its external or internal environment. BCI is direct communication pathway between an enhanced or wired brain and an external device.

The Brain-Computer Interfaces (BCI) project in Microsoft Research aims to enable BCI for the general population. This means non-intrusive methods, fewer number of electrodes and custom-designed signal picking devices. We go towards interactive BCI, which means response times within seconds and using EEG signals.

Activity of CNS

Direct measurement:

  • Electroencephalographic signals (EEG)
  • Functional Near Infrared Spectroscopy (fNIRS)
  • Magnetoencephalography (MEG)
  • Functional Magnetic Resonance Imaging (fMRI)
  • Positron Emission Tomography (PET)

Indirect indications:

  • heart rate, pupil dilation, galvanic skin resistance (GSR)
  • gaze dynamics, gesture/posture/gait dynamics

Neuroimaging modalities:

Recording method Abbr. SNR Temporal resolution Spatial resolution Probably portable Invasive
Electrocorticography ECoG High High High Yes Yes
Electroencephalography EEG Mid to low High Mid to low Yes No
Magnetoencephalography MEG Mid High Mid No No
Function MRI fMRI Mid Low High No No
Function Near-Infrared Spectroscopy fNIRS Low Low Mid Yes No

Electrical activity of the brain

Action potentials:

  • Single neuron electrical activity
  • Spikes 40 mV/1-2 ms/0-1,000 Hz

Local field potentials (LFP):

  • Group of neurons
  • 50-350 µV, up to 350 Hz

Electrocorticography (ECoG):

  • Electrodes on the surface of the brain (4-32)
  • 100 µV/200 Hz

Electroencephalography (EEG):

  • Electrodes on the skull (16-256): dry, gel, saline solution
  • 1-10 µV/50 Hz

Frequency bands:

Band Frequency, Hz
Delta < 4
Theta from 4 to 8
Alpha from 8 to 14
Beta > 14

Types of BCI

Passive BCI:

  • Monitoring the human state: emotion, attention, cognitive load

Interactive BCI:

  • Direct EEG decoding
  • Imaginary/inducted/stimulated movements, typically from the motor cortex
  • Attention decoding to audio or video
  • Evoked potentials: steady state video, or audio, or haptic, or …
  • Event related potentials: P300

Active BCI:

  • All the above
  • Induction of stimulae

Evoked potentials

One of the approaches in the interactive BCI
Types of evoked potentials:

  • Visual: steady state visual evoked potentials (SSVEP)
  • Audio: auditory steady state response (ASSR)
  • Haptic: steady-state somatosensory evoked potential (SSSEP)

BCI as type of HMI

BCI can be treated as another input modality of the human-machine interface. As such it should be used where it is more convenient or there are no other alternatives. Examples here are scenarios with augmented or virtual reality (AR/VR) glasses and hand-busy/eyes-busy situations. Such situation can arise on the manufacturing floor when hands are holding tools or are in protecting gloves (no gesture input) and it is too noisy (no voice input).

Our current research directions

We target general population (non-intrusive pickup with low number of electrodes) in interactive BCI (signal limited to EEG or MEG) scenarios. The most promising applications include augmenting the UI of AR/VR glasses with BCI components.