Appunti lezione Neuroengineering

PARETIC MUSCLE

Because of the reduction of fibers cross section and the conversion of fibers from type I to type II, the patient needs

a muscular training in order to increase the muscular force and volume and to increase the resistance to fatigue.

The current is increases progressively in order to check also the limit to pain.

9.3 T FES

HE NEURAL BASES OF FOR BRAIN PLASTICITY

We are interested in the effect of FES on the brain, we can see the contraction on the periphery, but rehabilitation

is about training an exercise with a backward effect on the plasticity of the brain.

CORTICAL LEVEL EFFECTS OF FES

1. (Neuromuscular electrical stimulation) NMES augmented voluntary activations increase cortical excitability

with respect to voluntary activations alone or passive NMES.

Transcranial Magnetic Stimulation (TMS) to check for learning properties of different groups of people with

different training paradigms. The TMS was combined with VOL, then TMS alone and volition VOL alone. When the

training was using both NMES and VOL there was an increase in the motor evoke potential, in the learning

process.

2. NMES combined with voluntary effort improves the prediction of sensory consequences of motor

commands.

The use of fMRI on healthy subjects allowed to compare the cortical activity induced by the 3 cases. NMES when

combined with VOL showed a higher cerebellar activity compared to the only NMES and a reduced bilateral

activity in the 2° somatosensory areas w.r.t the only VOL. Interpretation of results was improved by the use of

fMRI. Neuroplasticity is maximally optimized when FES is combined with VOL, the FES should be connected to the

intention of the subject, when available.

3. The NMES augmented proprioception in the context of volitional intent produced a higher activation than

NMES augmented proprioception in the absence of volitional movement.

We use a factorial design, when doing an experiment with multiple factors, there are different conditions that

could happen, we design the experiment to study all of them and we compare the conditions. The sequence of

conditions is randomized.

Here the FES can be on/off, and the VOL can be on/off as well, so the 4 conditions are: VOL+FES, only VOL, only

FES, passive (somebody is moving the articulation). We could see an activation for S1 and M1 for the ankle, which

are connected to the contrast: [(FES+VOL)-VOL]:[FES-PASS], which means: the FES in the context of Volitional

activity w.r.t Fes in a context of a passive activity.

➔ FES is effective when lower motor neuron are excitable, and neuromuscular junction and muscles are

healthy (post-stroke and SCI are ok).

The final goal is to study the physiology of healthy control. The main goal for patient study is to optimize the

selection of the therapy, the main question is whether the patient is benefitting from the therapy or not.

The clinical observation of FES is that we have the carry over enigma: it’s what was observed by Merletti in the

70’, he observed for the first-time using ankle-dorsiflexion stimulation for drop foot correction in post-stroke

individuals, that about 40% of chronic patients had positive outcomes; the other 60% didn’t.

At that time FES was patented for orthotic device for drop foot correction (for post stroke and Spinal Cord injured

patients). With post stroke chronic patient (stable condition after the accident, not still changing) they used FES

and 40% got carry over, the rest 60% got only the orthotic benefit. This means that, despite the objective was the

use of an orthotic device, patients had a recovery, a benefit, after quitting the training, they showed a relearning

effect. This effect is called carry over enigma, it’s an enigma because there are no clinical parameter allowing to

distinguish between the two groups of patients with different outcomes. For the rehabilitation point of view, it is

important to understand how to obtain this outcome. One possibility is to study what happens in the subject’s

brain to distinguish the two branches.

Patients undergo a pre-evaluation with functional tests FT, clinical scores CS and fMRI (2x2 factorial design FES +

VOL); after 1 month they are treated with FES for drop foot correction. Then, there is a post-evaluation equal to

the first one. The basal fMRI is processed splitting the patients into two groups: those with the carry over effect

(CE) and those without CE, to split them we use the results of FT and CS. To be consistence multiple clinicians

work on the same test. Once we have the two groups, we use fMRI to look for statistical differences between the

groups.

One of the results points out the difference in the activation in the contralateral supplementary motor area,

whose role is to plan the movements, healthy people have an activation of this area, and also the patients who

will have the carry over effect. Patients who won’t have the effect are almost not activating the area.

We can’t use this method on all patients, the problem with fMRI is that not all hospitals dispose the machine, and

the cost can be high (600€) since it’s an extra exam. We need to obtain this information with available device. We

need to check whether the cortex is involved in the movement planning. The requests for the devices are:

1. It must be a combination of FES + VOL.

2. Tasks need to be understood by the subject, the training should be included in a real-life task in order to have

a planning.

3. Neuroplasticity → we need repetition, learning comes out of practice.

In the case of drop foot, we are stimulating at the periphery the tibial-anterior, but it produces an effect in the

brain. The carry over enigma was associated to FES purely, the Merletti paper demonstrated the recovery capability

in patients who didn’t experience any recovery. According to Rushton (2003) the purely FES effects are associated

with the antidromic stimulus in motor nerves, the antidromic stimuli aren’t present in any other setting of rehab.

The hypothesis is that antidromic volley goes back to the anterior corn cells, interspinal neurons.

Interspinal neurons are the final step before the muscle activation, in case of a stroke the damage in the brain

produces a damage in the output signals going to the spine, signals are no more reaching the spine. Only a part of

the effective good signal still reaches the spinal neurons, together with a lot of non-functional signals conveyed to

the spine. What comes to the spine is a part of good residual signal which gives a good control on the task, and a

poor control signal (no control) that is noise. Then there is the output to muscles. Plasticity should be able to weight

more the good signal and to reduce the weights of noisy signals. That comes through training, so that the good

signal is the main information that effectively controls the output.

To increase the weight of a signal we use the Hebbian synapsis, the more they fire together the more they wire

together. If the antidromic volley associated to FES goes back timely paired with the good residual signal, it

increases the weight of the good residual signal. To have them timely paired, FES must be synchronous with the

intention of the subject, because the subject is the one producing the good residual signal. That’s what Rushton

proposed to obtain the carry over effect.

9.4 M NP

YOCONTROLLED

We need FES to be synchronized as much as possible to the intention. There are multiple sensor solutions to capture

the signal (EEG, electroneurograms, electromyography). The most applied solution is the Myocontrolled solution,

EMG controlled. If the subject still has contraction EMG it’s the clearest information, directly connected with the

task, and easy to read from the surface. There are two possible uses of EMG:

1. EMG-TRIGGERED NMES: trigger the target muscle to help the reaching of the task with a predetermined

stimulation. This kind of control assures that the subject initiates the task, but it doesn’t assure the

combination of FES and VOL.

2. EMG-CONTROLLED NMES, we read the EMG during the stimulation to check on the subject participation in

a close look modality. It modulates the stimulation parameters.

EMG-TRIGGERED NMES MYOCONTROLLED NMES

PROS EMG measures only before NMES Assure synchronization

starts

CONS No guarantee about Complex technological solution

synchronization

There is a difference in muscle contraction between natural volitional and artificial contraction.

• Stimulation artifact: spikes of few ms which provoke signals of higher values in the chain of the EMG reading.

• M-wave: synchronous contraction provoked by the external stimulus, huge value because all fibers are

synchronous, the good residual contraction comes after and it’s tiny.

• H-reflex second waveform.

• F-wave due to antidromic stimuli.

• Volitional EMG stochastic signal

Standard amplification unit for EMG recordings cannot be used in the presence of NMES:

• The stimulation artifact is the result of a potential difference produced by the stimulation current between

the EMG electrodes → it cannot be rejected by the differential amplifier.

• Since its amplitude is one to three orders greater than the M-wave, it can saturate or even damage the

amplifier of a standard EMG circuit.

• Different solutions have been proposed to face the problem of the suppression of the stimulation artifact.

Some possible solutions are:

• the blanking circuit, the circuit is synchronized with the stimulation trigger, and it disconnects physically the

EMG during the stimulation artifacts. But we still have high value.

• Low-gain amplifier, it prevents saturation and it’s still able to capture the high signal. With this amplifier to

read small changes we need a high-resolution ADC.

• Positioning the EMG transversal w.r.t the stimulation electrodes, so that the artifacts are more similar to the

signal, the differential part is only the signal.

• After the blanking signal, we still want to extract the right part also when immersed in the M-wave but getting

rid of the M-wave. Putting a threshold we would get the signal only at the end of the M-wave, not during the

M-wave. The main difference between M-wave and volitional components is the frequency. The M-wave

comes from the synchronous activation of all fibers, predicted, at low frequency; while the volitional comes

from the asynchronous activation of fibers, less predictable, high frequency → use of high pass filter.

• Adaptative filter: the M-wave is a predictable and repeatable signal, the output is fixed, we can exploit this

to split apart the volitional components which are the stochastic part of the signal. We get the EMG, we blank

the first part + we extract the volitional component as the difference between the EMG after the blocking

window and the average of the M previous inter-pulse recording, the average of the previous window gives

the predicted component of the M component. From the current window we extract the best estimation of

the EMG volitional.

PROPORTIONAL CONTROLLER

Theoretically it’s the bes