The following content has been provided by the University of Erlangen-Nürnberg.

Okay, good day everybody. Welcome to another lecture of biomedical signal analysis.

So last time we start event detection, talking about event detection, and today, we will

we will finish it. So let's first do the short test. So the first name, the first

question is name three waves or events that you can observe and analyze in EEG

signals. So do you remember something? Yes? Yes. So EEG rhythms. So you have alpha,

beta, delta, theta, and gamma. Then we talked also about some additional waves

and events. So you had like lambda wave, mirror rhythm.

And so on. We also say something about sleep spindle. And actually there is

like spike and wave complexes in epileptic seizure EEG. And actually today

we will talk about EEG event detection a little bit more.

Okay, the question two disappeared. So we'll go straight to the question three.

So do you remember the four steps of pantonkim algorithm? So what will be the

first step? Yeah? Yeah. And actually do you remember how we construct the band

pass filter? Yeah? Yeah. As cascade of the, so band pass will be the cascade of the

low pass filter and high pass filter. And actually how we construct high pass

filter? So subtracting low pass filter from the all pass filter. Okay. Then the

next step. Yeah? Yeah. So derivative based operator or differentiator.

Okay. Next step. Squaring. Yeah.

And the final important step is? Yeah, moving the integrator.

Okay. And after that it comes the searching method. But we said that

searching method of the palm and tomkin is not that good. Now you have a better

solution. And the fourth question is to draw schematically the output of the

moving window integrator of the pantonkim algorithm. So do you remember

what we said? So what will be the contribution of the P wave? So it will

be zero. So you will have no contribution from the P wave. You will have no

contribution from the P wave. And you will have some regular shape for QRS.

Does somebody remember what is that?

It's a trapezoid. And do you know the rising edge should be how long? Will it

last how long? QS complex. So you will have the rise edge that will be the same

length as the QS complex. So this will be the length of the QS. Then you will have

the flat part that actually has the duration of the window size minus QS.

And then the falling edge that's again the duration of the QS. And actually the

fifth questions, we didn't have the time to do to talk about it last time. So we

will do that today. Okay so just a brief recap. We first talked about the

definitions and the fundamentals of the event detection. Through some study cases

we talked about the motivation, why event detection is important, what you can,

which event you can extract from ECG, from carotidogram and from EEG. Then we

talked about some specific algorithms for QRS detection. So first we talked

about derivative base operator from the Balda that was published in 1977. So he

used the weighted combination of the first and the second derivative. And we

said that that algorithm was resource efficient but it wasn't that accurate

especially in cases with pathological bits. Then we talked about Pan-Thomkin

algorithm that actually had very good accuracy but it paid the price. So now

you are able also to detect the pathological bits but it actually pays

the price with the complexity. So the implementation is not that resource

efficient but still you can run it in real time on the microcontroller. And

these are the four steps that are important and after that as we said

comes the thresholding and the searching procedure which Pan and Thomkin

suggested and it was really complicated and not that resource