Brain-Computer Interfaces Flashcards
In which categories can motor neurons be categorized? How are the two categories defined?
There are upper and lower motor neurons.
Upper motor neurons include neurons from descending systems such as the motor cortex or the brainstem centers.
Lower motor neurons send their axons out of the brainstem or the spinal cord (e.g. in order to innervate skeletal muscles).
Name two additional structures, besides the Motor cortex and adjacent regions that are important to motor control.
The basal ganglia & the cerebellum.
How are lower motor neurons in the spinal cord arranged topographically?
All motor neurons that innervate a single muscle are grouped together into rod-shaped clusters.
Lower motor neurons that innervate proximal extremities are more medial within the ventral horn.
Lower motor neurons that innervate more distal extremities are located more laterally within the ventral horn.
Which two tracts of descendent motor control were discussed in the lecture on brain-computer interfaces?
The corticospinal and corticobulbar tracts.
Describe the pathway a non-Betz cell of the fifth cortical layer of M1 takes in the corticospinal tract.
Non-Betz pyramidal neurons found in the V layer of M1 descend in the corticobulbar and corticospinal tracts
They pass through the internal capsule, enter the cerebral peduncle at the base of the midbrain, pass through the base of the pons and coalesce to form the medullary pyramids
Near the caudal end of the medulla most fibers in the medullary pyramids are corticospinal axons
About 90% of these axons decussate on the height of the caudal medulla and form the lateral corticospinal tract
The other 10% of such axons do not change sides and form the ventral corticospinal tract
The lateral corticospinal tract forms direct pathways from the cortex to the spinal cord and terminates primarily in the lateral part of the ventral horn
Some of these axons synapse directly onto α motor neurons that directly govern distal extremities (mostly hand and forearm muscles)
Most of these axons however will terminate in pools of local circuit neurons that coordinate activities in lateral cell columns of the ventral horn
Define the term “brain-computer interface”. This does not have to be an exact definition, just be sure to include all key points.
„A BCI is a system that measures CNS activity and converts it into artificial output that replaces, restores, enhances, supplements, or improves natural CNS output and thereby changes the ongoing interactions between the CNS and its external or internal environment.“
Key points:
- measurement of CNS activity & conversion to artificial input
- replaces, restores, enhances, supplements, or improves CNS output
- changes ongoing interaction between CNS and its internal/external enviroment
Which four essential elements is a brain-computer interface made of? What is the fifth, non-essential element?
- signal acquisition
- feature extraction
- feature translation
- device output
(5. feedback)
What happens in the “signal acquisition” phase of a BCIs functioning?
Measurements of the neurophysiological state of the brain are conducted.
Such brain signals can be acquired through a particular sensor modality e.g. Scalp or intracranial electrodes, fMRI for Metabolic activity
The recording interface gathers neural information reflecting the intent, that is embedded within the brain activity.
What happens in the “feature extraction” phase of a BCIs functioning?
Digital (brain) signals are analyzed and pertinent signal characteristics (i.e. features) are distinguished from irrelevant signals.
What would a typical feature for the “feature extraction” look like, and what characteristics should it have?
Such features should have strong correlations with the BCI users intent
Typical features include amplitudes or latencies of event related potentials (e.g., P300), frequency power spectra (e.g., sensorimotor rhythms) and firing rates of individual cortical neurons
What is understood under the term feature translation?
Conversion of the pertinent feature into a device command using a translation algorithm
How does a translation algorithm work?
At the core of such translation algorithms is a model (e.g. a set of mathematical equations)
A model accepts the feature vector (i.e. a set of features) and produces an output
A model describes the relationship between the intent and the feature
Such description can be employed in order to convert future observations to appropriate output (generalization)
Do all BCIs require a separate feature translation algorithm?
No. BCIs using artificial neural networks can directly infer intentions from the acquired signal.
How are feature translation, device output & feedback linked?
The feature translation results in a direct command to an output device. Essentially it allows output to operate a task-specific device.
The device (e.g. a robotic arm) can alter its imminent environment and hence send feedback signals back to the BCI (e.g. breaking an egg, will result in us seeing that the egg is broken and result in sending signals to the BCI that then again reduces the pressure on the egg)
Explain the experimental design, that Pohlmeyer et al. deployed when researching the restoration of hand use in temporarily paralyzed monkeys with BCIs.
Two Rhesus Macaque monkeys had a 100-electrode array chronically implanted into the hand area of the motor cortex
Four forearm muscles were artificially paralyzed by inserting lidocaine or Bupivacaine in combination with epinephrine directly into the median and ulnar nerve
By extracting and translating features from the EMG signal measured in M1 intended muscle movement was predicted in real-time
This prediction was used to tune the strength of activation in the FES-electrodes of the forearm muscles
The height of the target (i.e. the force that had to be exerted) was systematically manipulated
20% of trials were catch trials in which FES was not applied
shortly summarize the results of the Pohlmeyer et al. 2009 study.
The nerve block decreased the amount of voluntarily produced wrist flexion in relation to the maximum voluntary contraction (MVC)
With the FES system activated the force exerted where significantly different from the forces exerted in a blocked nerve trial
Using this figure, explain how the BCIs function at second 1490 produces the results of this figure.
We can see here, that at s=1490 the BCI was turned off. Hence the flexion of the wrist did not suffice to reach the goal height of the target (grey square). However, if we take a look at the motor cortical discharge we can observe regular discharge patterns.
This overall pattern of results suggests, that the monkey did indeed try to exert force, but was unable to do so due to the temporary paralyzation of his radial nerve with lidocaine.
