Brain-Computer Interfaces

Question 1

What are the clinical uses of BCIs?

Answer 1

for severly paralyzed patients (e.g. ALS), BCI controls
- communication systems (e-mail, speech synthesis)
- media applications (internet or photo browser, music/TV)
- prostheses and wheel chairs
after stroke:
- BCI-guided therapy for faster rehabilitation

Question 2

BCIs can …

Answer 2

• recognize specific, well defined contro signals
• at least with some probability
• at least for some users
• speed, accuracy and robustness are currently rather limited

Question 3

BCIs acquire brain signals via …

Answer 3

• EEG
• ECoG
• intracranial microarrays

Question 4

What are the 2 kinds of BCIs?

Answer 4

• based on spontaneous oscillations (e.g. visual alpha rhythm) (intention, only binary behavior?)
• based on ERP (e.g. attention-modulated P300) (attention)

Question 5

What are “idle rhythms”?

Answer 5

brain rhythms that are attenuated during activation

Question 6

What is the trick when using the sensory-motor rhythm (SMR, around 14 Hz) in BCIs?

Answer 6

• modulation of SMR does not require actual movement, but also works with imagined limb movements (intentions)
• however, only 20-30% of population are not able to modulate SMR

Question 7

Why does Hex-o-Spell work better than Matrix Speller

Answer 7

Hex-o-Spell does not require gaze control

Question 8

What are challenges in real-time EEG decoding?

Answer 8

• spatial smearing: signal arrives at different electrodes with almost equal intensity due to different tissue conductivites potentially leading to high correlations between channels
• inter-subject variability
• inter-session variability
• inter-trial variability

Question 9

What is the Inverse Problem?

Answer 9

• find location of signal responsible for the measured data
• infinite number of solution
• assumptions required that constrain the solution space

Question 10

What are the matrices A and W regarding the Inverse Problem?

Answer 10

• A: mixing matrix describing propagation of sources to sensors, x(t) = A s(t) + n(t) (forward model)
• W: unmixing matrix extracting components/source signals from signal, multiple solutions (ICA, PCA), s(t) = W^T x(t) (backward model)

Question 11

Schultze-Kraft et al. (2015)

What is the readiness potential?

Answer 11

• slow, negative potential starting around 1 second before spontaneous, voluntary movement
  unanswered questions?
• Can a movement be cancelled after onset of RP?
• If so, is there a point of no return?

Question 12

Schultze-Kraft et al. (2015)

What were the possible trial outcomes?

Answer 12

• missed button press: participant pressed button and won, no light change
• predicted button press: light turned red before participants pressed the button
• aborted button press: light turned red early enough for participants to be able to abort initiated button press
• early cancellation: light turned red before any muscle activity

Question 13

Schultze-Kraft et al. (2015)

What are silent trials?

Answer 13

trials in which the BCI predicted a button press and time of the prediction was recorded but stop signal was not shown

Question 14

Schultze-Kraft et al. (2015)

Results

Answer 14

• movements can be cancelled after RP onset
• point of no return around 200 ms before movement onset