TASK 7 - fNIRS & BCI Flashcards
1
Q
fNIRS
A
= functional near-infrared spectroscopy = exploitation of different optic properties of oxygenated and deoxygenated haemoglobin
- oxygenated and deoxygenated haemoglobin have different light absorption spectra of near-infrared light
2
Q
fNIRS
- method
A
- place sources and detectors on the head surface
- -> send light via source in the brain - light spreads in banana-shaped pathways of photons
- light gets absorbed
- -> different amounts photons are absorbed - different absorption of light depending of the area
- -> changes in the Hb-concentrations causes changes in the reflected light intensities (active areas, absorb different light spectra/wavelengths than inactive –> because of the increased presence of oxygenated hemoglobin) - fNIRS measures alterations in the intensity of attenuated light at different wavelengths
- -> measure absorbance/reflectance changes at the 2 different wavelengths –> changes in the relative concentration of Hb and HbO2 can be calculated
- depth estimated by time-of-flight distributions
3
Q
method
- sources + detectors
A
- source = optode (e.g. LEDs)
- detector = receives reflected light; collect light after it has passed through tissue
- varying the distance changes the measurement depth
- -> distance = 4cm = fNRIS signal sensitive to hemodynamic changes within top 2-3mm of the cortex + extends laterally 1cm to either side
- -> the deeper, the weaker the signal (less sensitive)
- -> detectors ideally placed 2-7cm away from optode
- only measure distance between source and detector: can create blindspots –> need to consider which areas we want to measure to place source + detector correctly
4
Q
method
- Beer-Lambert law
A
= calculate changes in oxygenation from raw data
- applied when we measure raw data
5
Q
fNIRS
- hemodynamic response
A
- following stimulation/task performance, neuronal activity in brain region increases –> enhanced local oxygen consmuption
- within few seconds: blood flow, volume and oxygenation increase and deoxygenation decreases in area
- -> hemodynamic response - different optic properties of oxygenated and deoxygenated haemoglobin lead to the hemodynamic response being obtained by the fNIRS
6
Q
fNIRS
- optic properties of haemoglobin
A
- wavelengths between 700-900nm are used in fNIRS
- -> the wavelengths of Hb (deoxygenated) and HbO2 (oxygenated) change
7
Q
fNIRS
- HRF
A
- displays hemodynamic delay: BOLD response/reaction is recorded quite late + extended
- two graphs: 1 for oxygenated haemoglobin change and for deoxygenated change
- -> two dependent variables (in best case they are uncorrelated)
- -> gives additional info
8
Q
types of fNIRS
A
A. stationary
B. transportable
C. mobile
9
Q
fNIRS
- implementation
A
- time-resolves systems
- frequency domain systems
- above 2 provide info about both: phase + amplitude
- necessary for more precise quantification of fNRIS signals - continuous wave (CW) spectroscopy measurement
- apply light to tissue at constant amplitude –> measure attenuation of amplitude of the incident light
- somewhat less information than the other two but able to use LEDs rather than lasers (= safer), system is cheaper + can be designed to be small (= practical)
10
Q
fNIRS
- advantages
A
- non-invasive, safe, low-cost
- -> repeated measurements
- medium temporal resolution (sample rate up to 20Hz, but hemodynamic delay)
- no inverse problem (know where signal comes from)
- easy application, quite flexible placement of detectors
- noiselessness: less stressful, auditory studies
- no vulnerability to electromagnetic environment
- low sensitivity to head motion (more tasks, subjects)
- measure both: Hb & HbO2 (2 dependent variables)
- portable: provide continuous measurement, variable environmental conditions
- compatible with other methodologies –> compensate for pros and cons of another
- unique benefits to investigations of patients that are more severely impaired (e.g. psychiatric/ neurological conditions)
- versatile + customisable
11
Q
fNIRS
- disadvantages
A
- indirect measure
- middle-to-low spatial resolution: 1cm3 range
- limited depth pervasion
- low brain coverage
- light scattering by skin, tissue and other pigmented area
- between subject comparison difficult
- limited spatial standardisation: 10-20 system not optimal
12
Q
BCI
A
= system which connects human brain with computer
- allows for controlling external devices through brain activity –> without motor output
a) encoding a demand by features of measured signal - intentionally generate different brain states –> guiding subjects with sensory inputs (instructions or self-instructions)
b) decoding: translate encoded info back into original unit - differentiate brain activation resulting from higher-order processing or affective processing
- -> higher-order: attention, imagery; indirect info processing (= activity where we know that it can be differentiated well from another one)
- -> affective: automatic; direct info processing (= think what you want to say/do)
13
Q
BCI
- method
A
- measure brain activity (functional neuroimaging method)
- signal acquisition: extract specific features
- any informative aspect of signal that can be controlled by BCI user (e.g. activation level/ connectivity measure)
- -> digitalise signal - signal processing: translate these features into commands that operate a device (feature extraction)
- translation algorithm: rule-based algorithms
- -> machine learning algorithms improve accuracy of prediction/classification - command transferred to computer
- neurofeedback: overall task performance relayed back to individual as sensory feedback
- allows user to regulate state of specific brain function to achieve better performance
14
Q
BCI
- EEG-based
A
- use P300: computer detects P300 elicited when matrix elements containing the chosen alphabetic character are presented
- use slow cortical potentials + steady state visual-evoked potentials
- use motor-imagery-related synchronisation
15
Q
BCI
- MEG-based
A
- real-time BCI
- classify covert spatial attention (top/right/bottom/left to a fixation point) by using modulatory property of posterior alpha rhythms
- around 90% classification accuracy for motor + motor-imagery tasks
- better spatial resolution + better SNR than EEG
