Because when you... Very good, very good. Okay, so I should start again. But yeah, so...

All right, so yeah, briefly what we discussed was what I gave you into the... was basically

was written here, but I was mentioning that I gave you already the decomposed file,

and so these individual actions, individual firing discharge timings, you have one matrix that is

called motor unit pulses, which is this one. And basically this... Sorry, not this one.

This one. So you have motor unit pulses, and for example, motor unit pulses of the neuron number

one of the first firing, yeah, so this is basically the time sample when this neuron was active,

and you have it here, yeah, this is basically this line. So now the EMG I mentioned, it is

nothing but an algebraic summation of all the firings that comes from the spinal cord by the

muscle tissue. Okay, and the muscle tissues, it is basically a convolution of these axonal action

potential. So now you have in a file this data set, and this you can reconstruct because you

have the global EMG signal. So I already gave to you the deconvolved signal that you have to go back

with a convolution, and this is what we will do now. We will do this in this practical, but what I

was then saying at the end that we didn't mention before recording is that the observed experimental

EMG value has a noise component, and this noise component comes from the environment, it comes from

the other motor units that are not being decomposed, so it is just physiological noise and

environmental noise that you can imagine, and it's already something quite important to understand

that when you perform an averaging process, because this is what we do here, you take all

these action potential and you do a spike trigger average and you reconstruct one action potential,

you can already imagine that you are removing the noise from this data because you are averaging

a large number of action potential, and then you convolve this. You don't convolve the time instance

of firing, but you convolve the average of the time instance, and this is very important and it's

a very common, it's a very powerful method to test and to model synthetic versus experimental data.

It's very interesting because you can, for example, rule out if you can rule out if there is a lot of

contribution from the undecomposed motor units, for example, if there is how much residual there

is in the signal and so on. So, in the first task was to convolve the motor unit action potential

spike train. And this one, and this is exactly what you did, and so this is you did correctly.

So, I've seen that you've done this in a correct way, and then, and you plotted, perfect. Then the

second one was to generate EMGC, you know, so to obtain the clean EMGC, you know, sum the convolution

of the task one and plot the result together with the original signal, noted that you have to perform

this for all channels of the EMG grids and the corresponding one. All right, so before we go there,

just you see here, here you have all the neurons that were identified in this task,

and basically here you see that there is the action potential of that motor unit, that neuron.

So now, as you can imagine here, there are many action potentials because there are 64 EMG channels.

So, what did you do here, Raoul and Joana? Did you, and Nico too, because also Nico uploaded,

so I'm asking to you because I know you did it. Did you use the action potential from

one channel or from, or the biggest? What did you pick here? Which one did you plot here?

I chose the mean from the four maximum peak to peak.

Very good. This is a very good approach. Joana, which one did you choose?

I just choose arbitrarily one. So I just selected, I think it was channel one.

Okay.

I calculated the spike average for channel one and then I-

Okay, you just choose arbitrarily. Yeah, that's okay.

For each channel individually.

Yeah, perfect. Perfect. Yeah, so of course for this, we didn't give you any,

you were free to do whatever you wanted to do. So this is both solutions are okay.

If we would need, so it's quite, to find a good criterion is quite challenging,

but a good criterion would, sorry, also Nico did it. So Nico, how did you do it?

Nico?

So just had some problems, just got intense and against the goal.