I will be giving the next demonstration about P300 using the EEGLab and ERPLab analysis. So the previous tutorials, I have been giving based on the MMN and the auditory evoked potentials. Now, I will just give you an overview on background about what is this P300 about and how does the waveform look. And then, I will be giving a small detail about how does the demonstration being carried out. So it is a bit different from the MMN and AEP again in the event list how it is being created and some operations and all that is being done, how to visualise it and everything.
So, what is this P300? So, the P over here, it represents the positive peak that has been obtained and 300 means, it is like the positive response that has been obtained in the 300 millisecond that range. So, it will be obtained always in that 300, the 300 to 600 in that range it should be obtained. So, this is about the P300. So always, as I told you, there will be Wet Electrode. We have to keep -- it has been need to be put in the head and then there should be a contact between the electrode and the system, so that we can obtain the scalp potentials. So how is it, how the experiment has been conducted is that, there are basically three stimulus in this. So first is the standard stimulus that has been over there -- that has usually been given. And here we are, and P300 is about the visual, how it looks, how do we perceive whatever we see? That is how it is being recorded. So, there are two components that will be there in this P300. One is this P3a and the P3b. There are two components that is being obtained. So, one is the brain, it should be able to check how the brain is able to attend -- how it is being able to give attention to what we see. Like for example, now, if we have, there is a difference in the two, if we have a 4-centimetre ball and a 6-centimetre ball, we -- our brain should work in such a manner that it should able, it should be able to differentiate the 4-centimetre and the 6-centimetre. So, that is what is done by the attention.
That is, it is done using the attention how it would be, how we perceive it. So based on that, it is the -- how that is how the P3a works. And P3b is something that is based on the, how the memory is working. So, even if we have to differentiate between these two, we have to -- there will be another one stimulus also like this checkerboard or something like a distractor will be presented in P300. So, what will happen is, when this distractor has been given, we have to still remember that okay this 6-centimetre ball is the difference and 4-centimetre ball is also different. So, that difference also you should remember and we have to pay attention also and we have to keep it in the memory also. So, those two parts of the brain is being studied in detail using this P300 analysis. So, both the working memory as well as the attention-seeking memory of the cognitive brain, how it works, that has been explained or that can be studied in detail using the P300 analysis. So, here, as I told you there will be three stimulus basically. One is the standard, the standard it can be the normal and maybe in this experiment which we have conducted over here, we have used a 4-centimetre blue coloured ball as a standard.
So, and then, we will be giving a distractor as well as a target. So, these three are the stimulus that has been given. And about the experimental details, as I told, as I have mentioned, it is all the same, the neuroscan EEG system, where we have the 64 electrode channel. And then, we use -- we have, before acquiring the data itself, we have to have the BandPass filter, the Notch Filter and all the artifact rejections and all to be done has been performed. And the P300 mainly, it has been obtained from the frontal as well as in the temporal-parietal areas of the brain. So, in the next figure, I will just give you a brief about what are the different test stimulus and how does the, where does the, where do we get the brain, in which part of the brain we are actually obtaining these P300 responses I will be saying. So, here, this stimulus here, in P300 also there are various different types. Now, what we had performed is the visual P300, even there is an audio P300 also wherein, we have a target and its standard.
So, and there we do not have a distracter, but we will just have a standard target. So in visual, we have, a three stimuli, one is the standard which, like for example, if we have a 1000, totally we want 1000 stimulus mean over that the 800 stimuli will be over 1000, 800 stimuli will be of this 4-centimetre blue ball, the small sized ball. So that will be just like the responses which will give the N1, P2, P1, P2, N1 complex that particular thing it will give. It is just to sense; it is to just check whether our visual cortex are all working properly and all. So, that is for this standard. Next is this target. Target is actually a 6-centimetre blue ball. So here in this experiment, what we have to do is that we have to instruct the subject in order to give a response to the target. So, whenever there is a, this particular experiment is being running in that place, we will have a response pad been given to the subject. And whenever there is a target, that is the 6-centimetre blue ball whenever we are, whenever we encounter this ball, the 6-centimetre ball, we have to allow, ask them to give a response. So, that is about the experiment. And this, it will say that okay, the brain will differentiate the 4-centimetre ball and there is a 6-centimetre ball. And when there is a 6-centimetre ball, we have to give a response. So all of this, the brain will know and it will do it. So, that is why we say that the attention memory of the brain is being working and the attention as well as the working memory, both of them is being working in this condition. And there is a 10 percent of a Checkerboard also been given, like it is just a chess-board kind of thing with, this is how it has been given. So this will just, it is like acting as a distracter. Whenever, there is this Checkerboard comes again, we have to -- our brain will have to come to the normalcy that there is a standard and this target and now we have to differentiate again the 6-centimetre ball.
So, all of these, working memory and the attention, everything together, it works and then this is how we get the P300 response and in P300 itself, there are two responses, one is P3a and P3b. So P3a, it is like the earlier component and the P3b is a bit later component. Over here, if I show you and another one important thing about this is, it is all randomised. It is not like this MMN, always there should be three that beep beep beep boop, or beep beep beep boop like that, that is not the case in P300, it is always randomised. We have to give a percentage, like 80 percent I want the standards to be coming, 10 percent the targets and 10 percent distractor. So, whenever, it is like Artificial Intelligence or it is some program that is being done by the stimulus system, and it will give -- it gives in a randomised manner. So, in this condition for example, there are two, first two standards, then a target, then again three standards, then a deviant comes. So, it is all randomised. We want out of 1000 responses, out of 1000 stimuli, 800 will be for standards, 100 will be for the target and 100 will be for the distractor and it will be randomised. It is not that always, first five will be standard, then a target, then a distractor, not like that. It can be all randomised. That is the main important step stimulus thing that you have to remember about the visual P300. And if you see the image over here, the P3a, it comes in the frontal part of the brain and the P3b it is in the parietal part of the brain. So, frontal and parietal and mostly the central parietal, all these areas are the most commonly activated during the -- and the visual P300 stimuli or the ERPs are being obtained in these parts of the brain. So when we are doing the ERP analysis, also we should concentrate more on those parietal regions or the central parietal, temporal regions and all. So in that areas only the P300 response will be maximum.
So, there is a positive. As I told you, at 300-milliseconds, in that time, we will have this P3a, it starts the P300 response, it was obtained from that particular thing. And here we have a response being, as I told you, there is a response that should be given for the -- whenever there is a target being seen. So, these are the different stimuli and this is how it is actually being done and the experiment, I mean the brain, how it looks and how, from where it has been obtained. And then, the demonstration part is the similar as that. I only think that here, in this, we have to concentrate on the target and the distractor and then the P300 triggers based on that and even the epoching also will differ, because we wanted to be from minus 50 to like 600 or 650 like that, because our response in this case, it starts from 300 only. So, that is why we take minus 50 like a baseline and then to the next level like till 600 or 700 like that. We will be epoching in this condition. Similarly, all other steps are similar only, only the difference is in the event list and the epoching. And the plotting also, we can just give a difference or just give, how our, like for both the bin values we can directly do or we can do even for the separate analysis also.
And the P300, if you have to say about the papers that they have used, this Polich’s paper is the best paper for P300 because they are the ones, he is the one who actually started all the works about P300. So, he is the -- it is like a base paper for P300. So, as I told you, along with this visual, there are auditory stimulus also been given for P300. So, there are topographic analysis, everything can be studied with these papers. Next, I will be speaking about the demonstration using the Matlab using EEGLab and ERP Analysis. Thank you.
Hello everyone. So I was being discussed the P300 demonstration using EEGLAB and ERPLAB. So, I will just start with the demo with the Matlab. So, as usual, we just initialize the EEGLab and then here also I am using a CNT file obtained from the neuroscan system. So, I will show how the triggers look in this case. So, I just, first, we have to import the data. Here, I am using the CNT file itself. So, I just click this and then I have the P300 CNT file just open it, I will go with all the default parameters itself. So, here, so over here, we have totally, this is the P300 continuous file. So, here we have the 69 channels including all the trigger channels, all the other channels like the EMG, EEG and everything. Then, we have the epochs, there is no epoching done, so it is similar. Here, the events are 1000. So, as I said, here the 80 percent of these triggers would be standard, that is 800 of them will be standards, 100 of them will be targets and 100 of them will be distracters. And over here, see the time period, it starts, it is nearly about 1000 millisecond, 1000 or 50 millisecond. So, it takes at least 15 to 20 minutes for this particular experiment to be completed. And then it is huge, because it is having a lot of data, I mean, even the response is also being given so that is why it is a huge data size. Before that, before going to any other step, we have to do the channel locations.
So, here we have the channel locations. Here it is (Berra), the BESA file. It is like the default that is being given. So, it is always the 10-20 system according to that, all this FP1, FP2, CZ, PZ all the electrodes will be given in this and it’s like a default channel edit option. So, we will get all the 69 channels being red. So, it is just a default, we can just okay it. So, we have the channel locations been included. I will just give a small pre-filtering option of this, like I will just do a notch filter of 50 hertz. So, when we do that, our data will look better, good, because it is a raw data and everything, many artifacts and everything will be there. So, all of them will be eliminated. So, it is taking, it is due to a huge data size, it is taking some time for the data analysis, the filtering to be done. So, here, we have three data sets, three of them, three events are being there. So, we have, we can check that in the scroll data over here. So, now the filtering is being done. I will just show you how it looks in the scroll data. So these are all many artifacts and all are there, which we have to be removed.
So, I will just reduce the time frame. So for 30-seconds, I am showing. So over here, if you check, see these ones and all other standards, 2 is the target and over here this 3 is called as the distracter. So, the red ones, there will be 800 such red lines and 100 of these green and 100 of these violet lines. So, these are the different events that are being there. And actually, we can record the response that is given for this. So, as you see the target, the subject will give a response to this. So that response also can be recorded and been analyzed also, but we mainly want to look for these two and three triggers. So, next, I will just do, with the event analysis, event list. So just, we have to go and create the event list. So, I have already done the event list. So here the first one, it goes for the, the first trigger, it is a 4-centimeter blue ball and we just number it as a first bin and then it is a standard. The next one is the 6-centimeter blue ball, which is, which I had been put a bin number of 2 and then I do -- I keep the description as the target, because the 6-centimeter blue ball is the target and the checkerboard is the distractor. So, we name accordingly and number the bin values and the bin event codes accordingly and then apply it. So, what happens is all those ones, twos, threes and all will -- with B1, B2, B3 so it will be renamed in that manner.
So, here in this way, we have created the event list. So, now if you check the scroll data, you will find that all the things would have been changed. So, I will just change it to -- so, here all of them would be changed with the blue ball, checkerboard, the green, the 4-centimeter and all. So, it just differentiates these are the different stimuli that has been there like that. So, it changes, it has changed all the names. Next, we will be doing the epoching. So, as I have mentioned before, epoching means, we want to cut short a timeframe. Now, I do not want the whole of the continuous data for a particular trigger, I require only a required timeframe alone. So, for P300, it is in the, it starts from 300 only. So that is why I keep a range of about 700, 750 like that 700 maybe. So anyways, however it is, we have to keep a small baseline correction for any ERP. So, we keep a 10 percent of 700, it can be minus 70 or I am just keeping as minus 50 just to check how the waveform looks before the trigger is being started. So, we just run it and we have to, we will get a -- so there are 1,000 events means there will be 1,000 epochs that will be created. So, that is how it should be. So, here we have the dataset being obtained for after doing the bin paste epochs. So here, if you see, it has been changed to 1,000 epochs. Now, after doing these epochs, I will just again show you how it looks.
So, these are the each epochs. From this to the next one is called as 1 epoch. So, here we have I will just show you 1 epoch alone. So, it is from minus 50 to 700. So, this is one epoch. Similarly, we can get for all the bin values. So these are the first bins, we can go for even the others, even we can change the channel display also, but for now we do not require that and all. Okay. So, next after doing this epoching, next we have to work is the artifact rejection. So, here in this experiment, it is nearly for about 15 to 20 minutes. So obviously, there will be a lot of blinks and various other, various other artifacts like muscle artifacts, muscular movements or any neck movement anything can be there. So, here also we take a simple voltage threshold. So, all of these artifacts will be in the range of minus 100 to 100 microvolts alone. Here in this case, as I mentioned in the previous for the MMN, in that the trigger channels was not there, was included in that file itself. So, that is why we had to, we just selected the channels of our interest. In this case, we have the trigger channel separate. So we can take the whole set of all these channels, excluding the trigger and the EMG, EKG, these VEO channels and all. So, all these channels are the one which is due to the artifact, so that and all, we do not include in the artified deduction, because we want all these to be removed. So, that is why we do not select all these. So, we just select the 64 channels alone and accept it.
So here, if you check here, if you check in this. Yeah. So, here, totally there were 800, totally 800 standards will be there. Of that only these many of them have been rejected and this one these many have been accepted. So, this particular -- these numbers that has been accepted is the one which we average to obtain the ERP. So, this is the, this is the code that has been used for the artifact rejection code. So, here if you see these red lines, this red colour or the highlight ones, they are the one which is being rejected. So, we just update the marks over here and we reject it later. So, here in these tools, if you get the -- we can inspect this particular, we can just mark all that updated marks and we can just remove all of them. (Refer Slide Time: 11:02) So, as I showed you here, there are nearly 261 of them are being rejected. So from the main dataset, all those have to be rejected. So, here we just reject all of them. So, here all those 261 of them will be rejected and a new dataset will be, we rename it in this format. So, we will get a new dataset with the rejected, after rejecting all the artifacts. Then, after this also there is a small step of filtering again. So, again we do the filters, using the BandPass filter.
So, it has always the same from one. It is always the same from 1 to 30 that is always been kept. We can change, actually this yellow band gain and all is being coming, that is not, it is just that we have to change the low pass filter or we can change the order. But this is the default like for almost all the ERPs 1 to 30 will be the proper filter parameter, bandpass filter. So, we just do the filtering. So, it does for all the epochs continuously, so nearly 150, 750 epochs have been there for it. So it does the filtering for all of them and then finally we get the filtered dataset. So, here we have all the filtering done and only these many epochs have been there. Now these epochs are now further being averaged to get the different bins. So, here we have to do the average epoching. So, only the seventh dataset. We can actually take all the datasets also. But we do not want the filtered, or the ones with artifacts, the ones without the filtering and all, we do not require those dataset. So, we just take the single, this last dataset, which we have done with, which we have completed everything. So, we just run it and this is a small error about the desynchronization but doesn’t not matter.
Now, we just name it as P300 demo ERP. So, here as I do that, the ERP set will be here it comes. So, here this is the P300 data that has been obtained. So, next we can just plot all of this together. So, this is how it looks and we do not want these trigger channels as I said. So, these are something that is not required. So, here in this case, it is a P300 response, so that is why we want the positive to be up because the positive peaks will be obtained in this case. So we just take the default and we have from minus 50 to 700 we can actually change these values also. But if you take the autoscale itself, but then, it will give a proper, it gives a default itself. I will show the topography, so that we can see how it looks amongst the head and how it looks. So, here, there will be three bins will be obtained, one is for the standard, another one is for the distractor, another one is for the target. So, it takes some time for it to be done. So, for P300, usually it will be in this particular region, the most responses will be obtained like the parietal, central parietal like that. And if I, if you see that from the midline electrodes as you move from the most, I mean from the nasion to the anion, what happens is that there is a reversal that is taking place, all here it is all the negatives and here it is all the positives up. So, that is the difference. That is why, we say that the brain is acting like a magnet. So, for our interest,
So, this is the black line over here it is a standard. So, the standard, it actually gives, over here, it gives usually AEP, BEP responses and all it will give, even this target also will give. The main -- our P3a, P3b components starts from here, yes, this is the P300, the P300 starts in this particular part. So, this is the P300 response over here. The red colour is actually the target which gives the P3b that is the later part of it and this is the P3a that has been obtained. So, it is all coming and from 300 to in that range it is obtained. So, this is for a PZ. Similarly, we can obtain, we can get for the other responses also. So, if I say so this is the P300 response. If I consider, this is a P3a, if P3b, so here, see it is in the 5, it is in the range of 352 in this milli-seconds it is being obtained. So, this is a classic example of a P300 response that is being obtained. So, similarly, we can get for the other electrodes also. For example, even in the occipital also, the visual lobes and all the occipital lobes are the main place where the visual responses has been obtained. So we can even check-in that areas as well. So, basically this, if the parietal, central-parietal, so I will just show you some central-parietal for example this.
So, here we can get a proper response. So here this is the P3a and P3b that has been obtained. And then, if we have to check always from 250 plus, like that you have to obtain. So, this a proper AEP peak or BEP peak, that has been obtained over here. So, this is a classic example of all those P300 responses. Now, if we -- here, we can export also, this export option we can do. But what happens is we cannot export as separately based on the bins. Now, I cannot separate, separately the distractors alone or separately the target, separately the standards, together we cannot do that. So, we have to give the, while doing the event list itself, we have to separately give that while doing the event list we have to separately do the bins -- the events we have to select and accordingly we can do the preprocessing steps and then we can export the separate like the target separately, the distractor separately. So, everything can be separated only when we do this event list separately. In this case, we have given all the events we have been given as one event list itself. So, if we want to separate the targets and the distractors we have to do along like that. So, if you want, you have -- actually for P300 response, we mainly look only the target and distractors. So we can just select this target and distractors alone. And if you want, we do not want all the electrode channels. If you do not want them, you just want only the parietal electrodes, so we just take the parietal electrodes only. So, we can just take along like that. If you do not want the topography, you just want a straight line graph and that also you can do. So, it is always positive up. And if you want the standard errors of mean, also we can give so that you get accuracy of how the data looks and how variable our data, how accurate the data is, so that we can obtain.
So, here we can see a proper, the red one is the distractor and black one is the target. So, here we have the proper P300 response, that has been obtained in these, these are the peaks that has been obtained. So, here this, the parietal and mostly in parietal, the occipital parietal, in these areas only you will have a proper response. Since the CPZ is less because in the center, then the midpoint of the head, as I told you, there was a flipping happening, so that is why in the mid electrodes, it is becoming a bit constant or it is like very less. But it is there but it is a bit less that is it. So, here we can like this, if we can so this is how the P300 response looks like. And similarly, we can do the scripting also can be done in this way. So, we just have to put a command as EEGH, so that we will be getting the history of whatever we did. So, here we have all the, all of them which we did. So, first we loaded the CNT sample, then, the channel location which we added, then this filtering, then we plotted all those graphs to check how the triggers and all look like. So, all of them every, each and every step that we have been done, it will give you in this format. So, we can just copy this and paste in our other place, in our editor and then we can just change the subject, the name of the subject and then we can just change the ERP names and everything and we can obtain the different, for different, the multiple subject analysis, we cannot keep doing each and every subject the same procedure. So, for one subject if you do this analysis using this GUI and then if you obtain the script and then finally after that you can just, for many beginners and all can just do, doing this manner. If they are well-versed in EEG, in Matlab and all, where they can add loops and everything and finally they can do for multiple subjects they can do the analysis. So, that is about the P300 response.
Log in to save your progress and obtain a certificate in Alison’s free An Introduction to Brain-Computer Interfaces online course
Sign up to save your progress and obtain a certificate in Alison’s free An Introduction to Brain-Computer Interfaces online course
Please enter you email address and we will mail you a link to reset your password.