Skip to main content

Transforming the understanding
and treatment of mental illnesses.

Celebrating 75 Years! Learn More >>

 Archived Content

The National Institute of Mental Health archives materials that are over 4 years old and no longer being updated. The content on this page is provided for historical reference purposes only and may not reflect current knowledge or information.

NIMH Multimodal Brain Stimulation Speaker Series: Christian Windischberger, PhD and Faranak Farzan, PhD


Okay. Good afternoon. Thank you for your patience everyone in the room and online. Today we have two speakers, Christian Windischberger will be speaking now, and followed up by Farzan next, and so we'll be going for about two hours, and we'll have a brief break for questions in between. Please consult the chat window, which has links to the previous YouTube recordings to the other talks, as well as information on closed captions. So, I'd like to introduce Christian Windischberger. Thank you for coming.

Thank you very much for the invitation. All right. So the main focus of my talk will be transcranial magnetic stimulation. So TMS is a method which allows us to stimulate the cortex noninvasively. So we do this by using a coil up here, and this is because we have a very rapid switching electrical current, and this current causes a change in the magnetic field, and this causes activity in the (inaudible).

So, TMS is a non-invasive method because of the fact that the magnetic field declines quite rapidly with distance from the coil; therefore, we can only stimulate the cortical area. Importantly, TMS has become a promising therapeutic approach. It is actually FDA-approved as a treatment option in treatment-resistant MDD for the patient. So it is clinical application, therefore, of course, it is a very nice method for neuro scientific research, if you're interested, to stimulate certain areas in the brain using this non-invasive method.

We always have to keep in mind that the neuronal basis of TMS is not completely understood. So there are still some things there that we do not know all the details about, but, I mean, it's research, so maybe that's a good thing. So the target regions we basically have for TMS, the most important target region is always (inaudible) the primary mode (inaudible). So why is this the case? Because in the primary modes (inaudible) we get a very nice response, because we see the thing is moving. So when we have successful cortical stimulation we then see sort of movement of the thing or the area that we stimulate here. So this is a home (inaudible) for TMS.

Well, the point is most applications are actually interested in areas outside of the motor cortex, and in particular for the psychiatry application we're interested in the DLPFC, the dorsolateral prefrontal cortex. And here in this area, we have the problem that we do not have such an instant, instant sensor, instant instrument that allows us to find out whether or not we have successes to relationship or not.

So, the point is, in a standard TMS experiment, which will be focused in the motor area, one of the first steps that we will take in order to perform such an experiment will be trying to find out (inaudible) of threshold. And most threshold is the stimulation altitude that actually causes an observable movement of air or recreated in two recordings. We can find the recordings there. So, it's very important to understand that this approach is not available to us if we're interested in other areas of the brain, especially in the frontal and the prefrontal cortex. So we have to think about other options.

Now MRI, Magnetic Resonance Imaging, is a cool technique, and it allows us to get high-resolution images from the brain. Now, in addition to the standard MRI, we also have functional MRI, and with functional MRI we can also assess activation the maps in the brain. Now how can we now use MRI doctor, MRI methods, how can we use them and combine them with TMS? So how can TMS applications really benefit from these MRI?

Well, the first approach would be we can use MRI in order to do target selection, if we use a neuronavigational application. So we can do -- we can acquire an anatomy (inaudible), and this can be done (inaudible), and this can be, then, uploaded into the neuronavigation system. So when we aim for a certain structure as the target for our TMS experiment, we can actually -- in a real-time manner, we can actually see where in the brain we're going to stimulate.

The second approach will be a functional localized, and for this functional localized, we can basically use whatever function paradigm we are interested in. So we can use any (inaudible) paradigm that would evoke an activation pattern throughout the brain. And from this pattern we could then say this is the specific area that you would like to stimulate. So we could perform fMRI experiment, use the activation maps coming as a result of the fMRI experiment, and upload it into our neuronavigation system. And this combination would then allow us to target certain areas that have been actually activated by a certain target. So that's the first approach that we have, to combine MRI and TMS.

The second approach is we can have a look on brain activation or connectivity patterns before and after TMS. So this TMS application is sort of a repetitive (inaudible) TMS or rTMS, and this is the TMS version where we give a large number of pulses for a longer time. And this rTMS approach is the standard approach that we will use in psychiatry in the treatment of MDD patients. So we can use MRI as a powerful technique to assess brain connectivity patterns, and we can have a look on changes before and after an rTMS session on this brain connectivity pattern, or on the differences before and after TMS on brain activation maps.

Last but not least, let's see, so we've got the maximum that you can do concurrent TMS fMRI. So you have a combination of TMS and fMRI in the scanner, so we perform TMS, and we perform MRI at the same time. This approach is the most advanced approach, the most complicated, the most challenging approach, because, as you can see, bringing a magnetic field, a coil, TMS coil, which produces a magnetic field, a very strong magnetic field into a very strong magnetic field will definitely give some interactions between the two systems that they have to control for.

Now, nevertheless, this approach is actually allowing us to get some more information about the mechanism of TMS, because we have two independent methods, one method for activating the brain, which is TMS, and the second method for assessing brain activation pathways. So we can constantly assess the strength of brain activation depending on the kind of TMS stimulation that we have.

Now let me first start with the effect of rTMS. So rTMS, that's repetitive Trans Trainer Magnetic Stimulation. What These are the effects rTMS effects that have been reported in the literature? I'm just showing a few studies in there. So many studies have been performed in this area. So, first of all, in favor of Paus et al., study in 2001, they used (inaudible) rTMS stimulation and stimulated the dorsolateral prefrontal cortex, which is exactly the part of the brain -- the main target for TMS application in psychiatry. They found an increase in sort of - lots the DLPFC and the ACC.

In the Cho & Strafella paper in 2009, they found a change in dopaminergic activity in ACC and in the orbitofrontal cortex. And another important study in this region was in 2007, where the serotonergic activity in the cingulate and insular cortex, as well as the parahippocampal area was changed after performing rTMS.

Now if we have a look at rTMS (inaudible) surface, which is one study I'd like to present to you or show you a photograph here. This is one rTMS study, again, dorsolateral prefrontal cortex. And these are alterations in response at the stimulation sites. So, there have been attempts to use rTMS in combination with fMRI in order to (inaudible) more about the mechanisms that can affect the rTMS treatment.

Now we were also interested in finding out how does rTMS actually affect the clinical pathways in the brain. So what we did was we had 60,000 subjects, and we scanned them twice, in two sessions one week apart. One session was the actual stimulation session and the second session was the sham condition. So, within one session, what we have, we perform a resting state scan here, six-minute resting state scan in the scanner, then we applied the rTMS, and then we have two subsequent resting state scans after we've updated the subject again into the scan.

Now the idea of this design was basically to find out whether a single session of rTMS application will actually cause a change in connectivity pathways in those subjects. So that's the (inaudible) part, and there's also a sham part where we still needed to (inaudible). Everything else was exactly the same.

Now, we were thinking about a lot about ways to analyze this document in a way in which it allows us to minimize any bias in this scenario. And the idea that we came up with was we wanted to use a set of resting state networks that have been the result of a study in over a thousand subjects that have been published in 2010 by Bharat Biswal and colleagues. So, in the thousand-functional connectivity term project, they basically had a set of over a thousand resting state scans from many centers around the world, and they were trying for consistent resting state networks to be found in this very large data study. So, in other words, they wanted to find out are there some consistent state networks to be found across the world. And, indeed, they found about 20 of them. And these are the networks from the original publication.

So in our approach now to our rTMS study, we took use of these 20 resting state networks and used them as a basis, so this is an independent basis. It has nothing to do with our own study. We used them as a definition of the resting state networks. So analysis basically was (inaudible) that from all over 20 resting state networks found in this default study we calculated the main time post and correlated across all (inaudible). So doing that gives us a map like this. So these are the 20 resting state networks. And you can see these are distinct networks and a very nice example from what we saw from the Biswal study.

Now, the question was, which of these resting state networks is actually changed by using the -- by applying rTMS. Now every single network was then put into an (inaudible) and the (inaudible) contained all six resting state scans that we have from all these. So we have three scans in the varus position and three scans in the sham position. And we wanted to know, do we find any consistent changes in the resting state network across these six resting state scans. And if we do this, you can see as they drop out, all these blackout resting state networks did not show any significant changes across the different sessions.

But there's one resting state network that pops out. And if we have a closer look at this network, then we can actually see this is a network where we see actually part of that. So that means this r TMS stimulation appears -- gives us a specific effect only in the network with DLPFC is part of.

Now the question is what sort of changes we see in this network. Now, this is a network, again, if we have a comparison between the two resting state scans before and after DLPFC in relation, we can see that there is an increase in (inaudible) connectivity, right here in the ACC. So the ACC is always the usual suspect area that we are looking for in an EOP-assisting stimulation study. But I'd like to point out that, in this case, we found out this area with very little upper (inaudible) information was put into the analysis. So it was a really very highly unbiased way of analyzing stuff.

Now let's take a look in the maximum (inaudible) in there, have a look what's really changing in there. So these are the three resting state scans that we did in the real condition. Here is the actual rTMS application. This is the scan before, this is the scan after, and this is the new subsequent scan. So what you can nicely see is that there is quite a change in connectivity within interesting state network before and after stimulation. In the second resting state scan, you can see that this increase in connectivity is gone. Now as we have a look at the sham condition, we can see no change whatsoever. So this is a very, very specific effect that we find in there using rTMS on the dorsolateral prefrontal cortex.

Now if we now take this one box here, where we find the maximum change in function of connectivity and use it as a seed box for resting state, then what we find is this seed region here is connected -- in functional connectivity analysis, is connected very nicely to areas like the insula, nucleus accumbens, and the ACC. So, again, ACC always pops out in this sort of analysis in a way where we see that there is a change in ACC, and this change is caused by the rTMS application on the dorsolateral prefrontal cortex.

Now, the standard therapy in MDD patients is some pharmacotherapy. Now, we wanted to see what changes in patients in MDD patients, when they undergo all of them. So, in this second study we have a group of 30 patients that are suffering from acute MDD. They were either medication naive or they were at least medication-free for three months. The second group were remitted MDD patients. They were in stable remission, and also medication free, and 35 healthy age-match controls.

So the acute MDD patients, they received the standard pharmacotherapy as only applied in the Psychiatry Department at the Medical University of Vienna. Now this was an open label, flexible dose, unblinded but standardized antidepressant treatment. And we acquired two resting state fMRI sessions, one before, and one after the treatment, so three months apart. Now this, we also applied the same analysis, so we have a 20 resting state networks from the Biswal paper. Again, unbiased analysis approach, trying to find whether we see any significant differences in these networks. And, again, we find one network sticking out of those, and this is, again, the network number 17, which is the resting state network that includes also the dorsolateral prefrontal cortex.

Now here is the problem areas that's mentioned in one and two for the healthy controls, measurement one and two for the remitted patients, and the measurement one and two for the MDD patients. So, I should add that in this analysis we only included the respondents, just to make sure that it's a homogenous group that's in there.

And now if we take a look on the differences between the remitted patients, measurement two, and the respondents, then what we see, how we can see it somehow, that there is a change here in the ventromedial prefrontal cortex, so in gray you can see the resting state network, number 17 that we have before, and you can see a tiny spot here, where we see a significant change.

And this here is from the study one where we had 60 subjects in the rTMS and it's the same resting state network, and in red, you can see, again, the change that we found there. If we have a closer comparison of those two, you can see that we find areas indicating in both studies they are strikingly similar. So this, for us, is a very strong indicator that, indeed, the ACC is a very strong or important region effective in the MDD, patients and also an area that we can influence using rTMS. So this nicely fits together.

Now, switching now from these two studies to concurrent TMS/fMRI, so in concurrent TMS/fMRI, we really would like to acquire fMRI data at the same time as we perform TMS in the subject. Now this is the same as seven. So concurrent TMS/fMRIs have the potential to actually show what are the acute effects of TMS. So, in principle, it allows us to show us which regions are really activated by the TMS (inaudible).

The very first study using combined TMS/fMRI was published almost 20 years ago by Bohning, and for most applications, for combined TMS/fMRI, the standards have been the same. And this is the standard data, which basically is a birdcage coil. So a CP is a (inaudible) coil that, on the outside we have the TMS system right in there, with some sort of holder, which holds TMS in place, and the amount signal reception is done using this content. This is the standard approach.

Now these birdcage coils have a number of disadvantages. The most important disadvantage that we have on those birdcage coils is simply that the sensitivity is very low. About 10 years, 15 years ago, multichannel receiver ray had been introduced into MRI, and these are the kind of coils that are constantly used now in MRI, because they have a much higher sensitivity and they allow for us using a parallel acceleration method, parallel imaging, multiband imaging, simultaneous multisite imaging. So this is the state-of-the-art that we actually have in MRI.

In combined TMS/fMRI, we are, unfortunately, unable to use this acceleration technique, which means that the MRI techniques that we use for concurrent TMS/fMRI with this setup are actually outdated. They were used 15 years ago. Now our idea was, then, to find a way of coming up with a new way of acquiring the concurrent TMS/fMRI position, and this was to come up with a new coil. So we wanted to make -- have a tailor-made MRI coil that somehow circumvents all the disadvantage that we have with the birdcage approach. Now this made us want to design a multi-channel receiver rate, because only with a multi-channel receiver we'll be able to use neural imaging, multi-band simultaneous multi-site. And we wanted to have very high sensitivity right at the point where the stimulation takes place. So, for this reason, we needed to make sure that this MR coil goes, actually, under the TMS system. So we have the TMS system, and under the TMS system we mount the MR coil.

Now, as we said in the very beginning, distance is a very important fact for TMS. The magnetic field that is generated by the TMS system is declined rapidly with this system, so we need to make sure that the distance between the TMS system and the skull is not increased too much. So, in other words, we need to make sure that the cause of the field is very, very thin. That was not very easy to do. So, (inaudible), and, of course, the idea was we want to be more flexible. So, if we no longer need this birdcage coil outside, this means that we are much more flexible in where we actually find TMS, and this is really a champion in the applications that we can do.

Now, this means we have to think about the curvature of the coil to make it also possible to reach areas -- of the brain areas, and but the main application was still the dorsolateral prefrontal cortex. So in this region the overall diameter of the (inaudible). So this is -- again, this is the old set up. So we have the TMS coils inside. We have the birdcage coils outside. And if you compare it now to the new setup, where the TMS coil is in the same position as before, so this is the TMS coil here, and underneath the TMS coil is the RF coil. So it's mounted directly onto the TMS system, and this means that when we position the TMS system we automatically position our RF coil. We don't have to worry about putting it there. And, as you can see, the birdcage is gone, so we are much more flexible in which area you want to stimulate and which areas you are trying to reach using this approach.

Now it's a seven-channel receiver ray, so we said we wanted to have a mull-channel receiver ray in order to be able to use parallel imaging, multi-channel images, and so on. So this is a top view of the coil, where we have the seven channels in there, and the small things are the (inaudible). So this is actually an old set of the new ones. The interface box is much smaller. But this is the coil right here that goes under the TMS system. From here our cable goes to the interface box where all the (inaudible) are located, and from there this is the coil block that goes into the system.

So this was specifically designed for a three-tester system, and because our scanner that we used in our institute, our Siemens scanners that specifically build for Siemen scanner. And in the animation right here, you can basically see how the whole setup really looks like. So we take the direct coil, put it underneath the TMS system, and then we can go very close to the site of the point of stimulation and we get enough of advantages using this approach.

What's the most important advantage that we have is that we actually have the cone right next to the saturate (inaudible), and this means we have very high sensitivity. The second thing, multi-channel there, where we can use parallel imaging. We can use multi-level imaging in there. And the third advantage is simply it will be much more flexible, where we apply the coil and how we set up the whole experiment.

Now, this is the magnetic field, the (inaudible) magnetic coil. The TMS system is up here, and this is where the TMS system is, again, up here, but in addition, we have our RF coil put in there. What we see is there is actually no change in the overall pattern of the magnetic field that is produced by the coil, by the TMS system, but we see there's a slight reduction. And this reduction is simply due to the fact that the distance is now increased, the distance between the TMS system and the cortex, in the case, of the test object is increased. So we have a loss of about 15% of the TMS stimulation amplitude, which means we have to increase the amplitude on our TMS intensifier by 15%, approximately.

Here, you can see the outline of the TMS system, and this is the position of the coils underneath there. So, we have to manipulate the frequencies of the coils because they're so close to this big TMS system in such a way that all of them are optimally matched and tuned so that, again, increases our sensitivity to the whole system.

Now, a secondary very important issue if you want to use those kind of coils for parallel imaging, is we need to quantify the G-factors. Parallel imaging is a way for us to understand (inaudible) and speed up its position. Now the principal approach that's used in parallel imaging means that if you use parallel imaging you have some reduction in sensitivity. And this is an automatic feature when we use parallel imaging. However, there's a second feature in there, which, in addition, reduces (inaudible) ratio as soon as you use that (inaudible), and there's a simple G-factor.

And, nicely, what you can see is the G-factor here is almost one, so there's no additional reduction in SNR, if we use a parallel imaging factor of two, a factor of two, percent factor of two. So you can see these two factors are homogenous, and it's close to one, which means this is perfectly suited to perform imaging using parallel position with a factor of two. And if you go to higher parallel imaging factors, you can see that the higher values of G here occur, so three might not be a very effective, but two can be done. And if we use a parallel imaging factor of two, it means we reduce the acquisition factor by 50%, so we're much faster (inaudible).

Now, the second question is, what about sensitivity? And here is a comparison of the signature noise ratio of this new coil compared to the birdcage coil. Now you can see here the RF coil was here, mounted in the position at the occipital lobe underneath the patient, and the further we go away from the coil the less sensitivity we have. So we have the maximum sensitivity boost in areas that are very close to the coil. And this line here, this black line here, indicates a five-fold increase in SNR by using the new coil compared to a standard birdcage coil. And even in deeper areas of the brain here, at this step here, we have an equal SNR using the new coil compared to the standard coil.

But the main idea of the coil was basically to maximize the sensitivity right at the point where we stimulate, because one of our main intentions was, we wanted to see what happens in the brain when we actually use TMS. And here is what our old system looks like. This is the TMS system, and underneath here is the coil.

Now, this is a comparison. This isn't a comparison. This is just an example of how the images would look like. So here you would have the example we see, and, of course, you get very nice images from them. You get an increase in the intensity in there, but actually, even on this image, you could see what happens in the front layers here. It's just because of the scanning that it (inaudible) like that.

Now, if we use a TMS system inside an MR machine and right next to MRI position, MRI imaging position, we have to make sure that there are no interactions between (inaudible). So what we tried was we would take a look at EPI images. So this is an EPI image on this slide, and we tried to see what happens in EPI images now, depending on the time between the TMS pulse and the EPI position. And you can see that if we have up here, so which is just about five milliseconds, or so, difference between EPI and TMS, you can see there is a whole blackout of the images there. This is not a very big surprise, because, as you know, in MRI what we do is we use frequencies or magnetic fields in order to code information in there.

When we use TMS we also change the magnetic field, and, therefore, these two don't go well together. If we have a difference or a distance temporal distance of maybe at least 50 milliseconds there, you can see that there is no effect on EPIs whatsoever. So this is basically the interaction that you have to look -- that you have to be aware of between the TMS application and the EPI application.

Now, in the comparison of the slices without TMS and this is using a concurrent TMS approach. And you can see the quality of the images is almost identical irrespective whether we use TMS there or not. So that's a very good result, because that basically tells us we can use this setup in order to perform (inaudible).

Now, the very first question, of course, is does whole setup work? What we tried was a simple block design sort of thing. So the yellow stripes here are the imaging acquisition periods, and the blue blocks here are the blocks where we actually applied TMS in there. Now if we zoom into these block, this is the EPI acquisition, and these are the TMS blocks. So we have EPI acquisition and we stimulate acquisition as we stimulate (inaudible). So it uses in these block-wide fashion, and within one of these blocks in there we have eight pulses of 10 hertz that we apply to the subject.

Now, does it really work? Yes, it does? So that's the result of the first subject right into that, the same subject that is currently speaking to you, so it works. You can see in the hand area, in the primary (inaudible) cortex, this is kind of response that we get from exactly those blocks in there, so the principle works. We do see activation caused by TMS.

Now, this is from the original (inaudible) set, so this kind of setup, where we actually have the RF coil underneath the TMS system, really allows us to get this very high sensitivity. And for that reason, we have, here, strong signal changes there. You can see it's about 3% signal changes we find in the head area of the primary (inaudible) of the cortex. Not caused by some hand movement, really caused by performing TMS, applying TMS onto primary.

Now, the next interesting question is, is it somehow possible to assess how much dose response we get with different TMS stimulation altitude? So, one of the first things that you would do if you would have a new pharmaceutical treatment for (inaudible), is you would need to find out what's the appropriate dose. And this dose estimation in TMS is not so simple, because for most of the time the way we assess dosage there -- well, we use the multi-threshold as the main calibration around. However, it's easily possible that the multi-threshold, which is just defined in the primary multi-cortex, might be insufficient in order to assess the kind of stimulation activity that we actually need in other areas of the brain, in particular the dorsolateral prefrontal cortex.

So we wanted to find out whether it's possible to use this concurrent TMS/fMRI in such a way that we can actually come up with the dose response curve. Now, so we wanted to know whether this fMRI can be used for that? And home ground for TMS is always the most important, so we tried this in primary multi-cortex. We had seven healthy subjects, and we performed two sessions, two MR sessions with them.

So, in the first session we'll be using the 32-channel head coil, which is the standard head coil that we were always using with the system, and in the second session we were then using a 7-channel coil in combination with TMS, for real concurrent TMS. This study was done with three Tesla, made for three Tesla, and with using the multi-band sequence from (inaudible). Temporal resolution here was about one second. We acquired 14 slices, and (inaudible) resolutions of 1.5 by 1.5 resolution, a thickness of about 3 millimeters.

Now, as I said, we had two sessions with the subject. In the first session, this was actually more a localized session, so we were using the 32-channel head coil. And in the session, we acquired an (inaudible). This was important because we needed it for neuronavigation purposes. The second think was we performed a short finger-tapping paradigm to find out where the primary modal area is, where hand area for the primary mode area is located. We used FT for targeting TMS using neuro navigation. We used this information -- we used this first session also to get some information about the coil positioning, so where the coil is actually positioned in there. And we used it to determine the motor threshold. So this was all done within this first session.

Then we took the subject out, changed the setup, so from the 32-channel coil to the 7-channel coil. Again, the coil (inaudible), which we can (inaudible) co-register to the first dataset. Again, perform a finger-tapping task and applying TMS. Applying concurrent TMS (inaudible). So this was the paradigm that was used, so you can see, also, this current TMS right here, one round, and this round was about 7 minutes long, and, yet, the blocks of different TMS stimulation altitude. So we have 80, 90, and 110 percent of the individual -- of the threshold, and we wanted to know whether or not we see a change of the whole activity in the (inaudible) area with those different stimulation altitudes.

With this neural navigation, in order to accomplish this task, so this is, again, the center, where we have the subject here, you have the coil here, and the TMS coil, and then you need to come up with some sort of approach to allow for your navigation there. So most neuronavigation systems are incompatible with the MR environment, so what we did was we used what we have and got rid of all the (inaudible) objects or all (inaudible) magnetic objects in this setup, so, therefore, we have this sort of wooden stand here where we mount the camera there, and this camera is able to get the tracker. So we have one tracker on the TMS coil, and we have a second tracker on the subject. Using the two together, then allows us on the neuronavigation system to actually see where we are targeting. And we can change the position of the TMS system such that we actually target the one reach we're really interested in. And in this case, this reach was to find the thing in there. So, with TMS applications you can actually use your standard neuronavigation system, also to be used in the vicinity of the scan.

Now, here is a comparison of image quality. Now acquired with this center channel head clamp in a plane that are parallel to the plane defined by a coil. But this is an EPI, and this is an (inaudible). So that's an (inaudible). That's an EPI (inaudible). What you should be able to see, it's much nicer on the left up here, but you should see that, actually, the structure that we see on the top, opposite here, that field is represented in the EPI there. So we have a very nice congruence of those two [inaudible] acquired with different modalities in there. This is very important because it basically allows us to use the anatomical opposite as sort of a basis for finding out wherever you want to stimulate.

And here are the results. This is one subject. This is the result, activation back from the finger tapping paradigm in all three planes. This is the activation that's been seen when we use 110% of the individual's threshold and 100% of the individual's threshold. Global maximum and we also took a look at the local maximum right next to the hand area. What you can see here is that there is a difference in the extent of the activity between 110% and 100% percent of both the threshold simulations, which is a regular indicator that indeed we can try to -- We can use the approach for dose assessment.

Now if we extract the parameter estimate in these areas and average across the group, then we see something like this. This is for the voxel of the global maximum activation for 80, 90, 100, and 110% of individual threshold and you nicely see this monotonic increase in the activation altitude. Now if we have a look at the voxel right next to the hand area, we can also see this increase in activation altitude. Only if we use 100% individual threshold we get a statistical significant difference. So, this simply shows you that the approach of combining this seven-channel coil with the [inaudible] system you see enough sensitivity to actual come up with [inaudible]. Now in multi-cortex, we don't really need it. Because in the multi-cortex, we have an immediate output parameter, which is whether we see movement or not. But we can use this approach in other areas of the brain. This is the next step we want you to do.

Now, we wanted to apply dissimulation on those lateral prefrontal cortex, the left lateral prefrontal cortex. However, we are also interested in seeing the local effect, but also the seeing distance effect. Now, the one disadvantage that we have with the usage of the seven-channel coil is that the -- We are limited. The sensitivity increase is limited to a certain distance of the coil, so the question is -- While this approach alone will not give us whole brain current, so what can we do? So if one coil gives you such a coverage there, well, we can just add a second coil and increase our coverage by the second channel.

Now this is the new set of them with two coils, where we have a very happy subject. You can see the TMS system right here. This is the TMS system. This is one MR coil, which is mounted again directly underneath the TMS system. This is the second MR Coil, which is positioned semi-contralateral to the first. Now, this approach has a number of advantages. One of the most important advantages is if you're interested in using [inaudible] simultaneous multi-slicing imaging technique because this approach allows you to go to very high multiple factors without a significant decrease in the [inaudible]. Now, using this approach -- We have 14 coils in there -- actually enables us to do very rapid acquisition of the whole brain. This is very dark there but I hope you can see -- You can actually see the whole brain using this approach with the two coils there. The yellow line up here indicates the position of the TMS system, so we actually know where we are.

We enter the local and distance effect, so we use the TMS coil over the natural pre-frontal cortex and position a second MRI coil contrary to this. We were performing two different runs in here. One run enables about -- So we used 1 HZ concurrent TMS/MRI and it looks like this. This is the EPI position here and then we just use one pulse; then the next EPI position and another pulse. So there's one second delay between those 2 pulses, so there's 1 HZ stimulation [inaudible] in there. This all takes part within one TMS block and we have a number of the TMS blocks with different TMS simulations altitudes.

A second approach, we were also interested in finding out differences between 1 Hz and 10 Hz stimulation there. In this case, what we have here, this is the EPI position block and then we give three pulses with three [inaudible] right in there and then we have the next EPI position. So that's the paradigm that was used in this study. Importantly, every resolution is one second and because of the use of high [inaudible] factors, we actually were able to come up with 28 slices and have rather high [crosstalk] resolution. Each resolution is 1.5 [inaudible].


Identical to what we use in the voxel studies were 80, 90, 100% and 110% individual threshold.

This is the results for the first subject. The green spots here indicates where we actually stimulated and what you can see is not enough. We see barely activity in there. So seeing these results were disappointing, so the whole thing did not work. The only areas reacted are acoustic areas, so it was loud in there. The TMS noise was definitely there. The first idea was that TMS was not working, but TMS noise was there, so what went wrong in there?

Now -- This is a comparison of the anatomy and position of the TMS coils, so I hope you can see somehow. We have two colors in here; green and red. I should mention first that in order to find out where exactly the position of the TMS coil is, we imparted or we mounted in the seven-channel coils three small spheres that light up in MRI image. They allow us actually to define a plane of the coil and position the other slices accordingly. And in green, you can see the position of the TMS system before we actually run the TMS and in red you can see the position afterwards. And what you'll also see that's significant change in these positions. So in other words the subject moved away from the TMS color and moving you away from the TMS color automatically means that the distance between the color and the cortex is increased and if distance is increased then stimulation intensity that we see on the cortex is decreased.

This is also very nicely seen here. This is plain before and after TMS and you can actually see the shift in there. So in this case because the subject is shifting away from the color we do not see any activity. Now let us look on a different subject. And here you can again see the head up here before and after and I hope you can see that they're almost identical. So this is a good subject, the subject didn't move a lot before and after and if we have a look at the activation maps that we see in the subject, they are much, much nicer.

Here is the area that we actually stimulate and you can see by stimulating TMS simulating this area of the prefrontal cortex, we get widespread increase of brain activity in a well-defined metric. So if we really stimulate just on the left dorsolateral prefrontal cortex, we also stimulate, we also not only increase the activation right there where we stimulate but we also see activity on the contra lateral side and we see activity in areas that are well-connected to this particular region. We can make the Ganglia, Thalamus, all sorts of regions actually light up. And this is a very strong indicator for using TMS as a real tool to stimulate the brain. And not only stimulate a single area, a single focal point, but actually using TMS you can start to stimulate the whole network that's connected to this stimulation point.

Here is what happens with [inaudible]. Again, I'll just switch back and forth so you can see the difference. So [inaudible] did not change but the frequency of the [inaudible] did change. And we see quite a different pattern of how brain activity is now lighting up in this [inaudible]. So this also indicates that frequency is an important parameter in our application. We still don't really know what the optimal frequency will be, but we see that frequency has quite an impact on the pattern of the distribution of the effects there.

Now here's the direct comparison, so one has stimulation, which has always thought to be more or less inhibitory. Here's what we see when they - we see actually higher activity on the contralateral side there, we see widespread activity also in the basal ganglia and this is the activation map that we find in Penert's in his experiment. Again, here is where we stimulate and we see strong activity on the contralateral side there. I'd like to find out if this single subject lights up. So these are really things that you see in a single subject, they don't [inaudible] this is what you get in a single subject.

Now we actually wanted to know something about the dose response curve so here is the results for those responses in the dorsolateral prefrontal cortex. So this is our target area, the DOPOC and here is the response at baseline. So this is for 18th, 19th, 100th, and 110th of the individual with the threshold. And what you can also see is there's an increase in the response there. Now what happens on the contralateral side we can also see an increase from going from low stimulation amplitudes to going to very high stimulation amplitudes. I think a very nice result because it actually shows you that indeed by selecting your stimulation amplitude you change something, there's an effect in the brain depending on how much is stimulated.

We also have activity somewhere here in the ACC area of the ventral medial dorsal area here and there is something quite interesting, well we found it quite interesting. Whereas here from the site of the DOPOC where we actually stimulate we see it's more or less more of a tonic increase and we more or less see the same thing on the contralateral side, this area here shows quite different patterns there. This could be caused by the fact that the ACC here is not really directly connected it has a much more role of modulating connectivity. So modulating brain activity there. So not such a direct connection that it really has as on the contralateral side. And then we see a different pattern in there. Again, this is the same subject this is the same subject that we already have. This is from the very first study that I showed you on the other 60 subjects without TMS and again we find a usual suspect in there. We find again the ATC area is an area that's somewhat implicated in the whole stimulation process.

Right and finally what we also trying to find out what's the functional connectivity network so we have this seat now at the position where we actually stimulate using TMS and this is the next thought that comes up. This is where we stimulate and in addition we also acquire [inaudible] in this subject and you can really nicely see that we find strong connections between the two areas of the hyper stimulation and the activity on the contralateral side.

So taken together, very strong indicator that using TMS we can actually stimulate-- we can not only stimulate a single cortical area but we also stimulate all the areas the whole [inaudible] that are connected to this cortical area. So concluding or trying to [inaudible] all of the potentials of TMS FMRI potential in [inaudible] work the cool thing with TMS is that we can assess causality. That's something that's not really possible with FMRI alone so combining TMS and FMRI allows us to introduce causality. We know that it's been used and has been used in clinical treatment but we have the problem that applications around this are quite heterogeneous. Sometimes have certain rules of thumb character and are I would say the foundations follow from the choice parameters seems somehow limited.

Now the desired location and the [inaudible] across studies so we have some people that have a 5 centimeter rule when they want to find the dorsolateral prefrontal cortex, some say use the four finger rule that's much better to find that. So that's quite different in there and we also have to think about the frequency and the amplitude that we should use in order to stimulate. The diagram that I showed you I think nicely shows that you can use concurrent TMS FMRI to find out the optimal stimulation amplitude. The amplitude actually guarantees an effect in your target area, which is something that I'm able to do with without these sort of methods.

These studies [inaudible] allows you for assess TMS [inaudible] spike of stimulation. If you combine with the second of these [inaudible] this enables you to increase the coverage of the whole brain. Using a [inaudible] array allows you to use parallel imaging [inaudible] and it also allows you to use [inaudible] approach. And taken together, these methods speed up your position considerably and actually get you in the same position as standard MRI [inaudible]. Now the increase in sensitivity allows you now to actually see happens at the site of simulation. In the old setup, where we have the birdcage coil outside and the TMS coil inside, the problem was that this TMS coil was sort of shielding local -- or was reducing the sensitivity of the site of stimulation.

With the new approach, you have, actually, the highest sensitivity right by your brain stem.

So this setup is ideal to test differences in BOLD activation across different stimulation parameters.

Now, the whole setup now, in my view, is bringing the application of TMS into a more precision medicine sort of approach. So, with the idea having that for every subject, for every patient, we could determine the optimum stimulation amplitude, the optimum stimulation location and take this information, then, for the actual application of rTMS to the subject.

This allows us to really ensure that the TMS has effects in the -- has the intended effects in the intended cortical sites. We also see that if we successfully stimulate the cortex, then we stimulate a whole network that's connected to this cortical area.

And still, there is a lot of work to do in order to come up with the optimal parameters, for the parameters that are really ideal for patient outcome. It might be that just -- I don't know. One, two, three sessions of rTMS might be sufficient for patients to improve. It might be that we need to apply the rTMS every four weeks, but maybe just a single session. So these are all things that are unclear, that need to be determined. So there's lots of work to do there in rTMS of the future.

And of course, I'd like to acknowledge a lot of people who have been working. So that is my group, and I'd like to acknowledge, in particular, Martin Tik, who spent a lot of time in optimizing TMS, fMRI, and optimizing this setup and making sure that things are really working fine. Lucia Navarro de Lara, who actually built the TMS coil, and Michael Woletz, who look care of the controlling of the TMS simulator, and the rest of them.

With that -- And of course, I'd like to acknowledge my collaboration partners from the -- collaboration partners from the Psychology. So that's Claus Lamm, Markus R�tgen, Daniela Pfabigan. And from the Department of Psychiatry and Psychotherapy, Rupert Lanzenberger, Andreas Hahn, and Georg Kranz.

With that, thank you very much for your attendance.

So I think there's time for maybe just a few questions in between talks. If anybody in the room has questions, just raise your hand and use the microphone. And if people online have questions, please use the chat window, and we'll read off your question.

All right. Great talk. It was very inspirational.

Just a quick question so everyone knows. How can someone get involved in this research? Is this a commercial product? How can they start doing this themselves?

It was very important to make sure that the theoretical aspect can actually be easily applied to patients. So therefore, we tried to make sure that it is C-certified. You know that to get C-certification, you have to collaborate with a company. We collaborated with Magventure. So, the coil is specifically made to fit nicely under the Magventure, a non-comparable TMS column. And it's also distributed by Magventure.

Hi, very nice talk. Thank you. I have a couple of questions, actually. One begins with the very basic things like multi-threshold. If you measured multi-threshold with this coil before or after, you were just mentioning applying an aMT 100% of a threshold, of the primary.

So we determined the threshold outside the MR scanner first. Actually outside of the MR scanner room. You think E and T electrodes, and we then just barely fired, and the multi-threshold when the subject was in the scanner. This is because we have a very long cable that's connected to the MR compatible coil to the TMS system, and therefore the stimulation average was quite different. So, therefore we calculated what should be the optimum stimulation averages and varied with the sizes of the cable.

Okay, great. Thank you. Now, coming back to the last result you presented, very interesting with stimulation of possibly the frontal cortex. Could it be that the [arterio-singular] activations, or actually, the activation that defied with higher averages just because of the pain? How was it perceived by the subjects? This kind of stimulation at higher percentages?

It was not perceived as painful. What we found was that because TMS causes moving when they're under stimulation inside the magnetic field, it causes pain. It took us a lot of time and a lot of effort, and Martin Tik spending weeks on that, in the wake of finding an option so it's comfortable for the subject to have the coil being attached to them, and still not increasing the distance too much between the coil and the scanner. So, what we basically had is we had some sort of a cushion that allowed you to reduce this sort of head-banging.

Okay, and the last question is, can you speculate on why you have an under-activation for lower percentages? The 90% was definitely more negative than what we had tested with the others.

That doesn't really answer the question. I wouldn't really want to speculate about that. The really important thing is that we see that the change across the different stimulation entities. The activation might also have something to do with what's baseline in this experiment in them. It could also be -- We need to think about whether each percent, what threshold, it's simply not enough to induce anything in this cortical area. It could be that if we used that -- Let me rephrase it.

I think one of the main reasons why we see a rather low response, well, not a very high response rate, in patients when they are let go without treatment. In many patients, we might just not stimulate enough. For this reason, we don't really know what happens in this really very rather low stimulation amplitude. It could be that we are not actually stimulating the prefrontal cortex where we really focus, but something, some area that's closer to the coil. If it's closer to the coil, then the stimulation would be higher, so the magnetic change would be higher, and then we might just see a sort of collateral defect and it's not actually the prefrontal cortex.

Many people contrast rest and stimulation, right? This is true, but if it's all in one go in there, I'm not so sure what the reason really is for this, why we see a deactivation there.

Could that perhaps be related to activating different populations of neurons? If you're going from numerous thresholds?

It could be.

Speculating from the experimental protocol in which you stimulate [inaudible] for any reason to activate a different population of neurons?

Actually, I'm not going to have to speculate too much about that because I think most of the results, most of the features about TMS that we know of have been measured in primary motor cortex. And I do believe that things in other areas of the brain are quite different. So, yes it would be a very good explanation for it. Some different population of neurons that are kind of getting activated with the TMS, but I think at the moment the evidence is not there to actually let us see if there's any indication of that.

Have you compared this at all, in those individual subjects with TMS, EEG results?

No, we haven't. So, of course it would be really nice to have EMS, fMRI, EEG, in one go, but I think there are significant challenges in TMS fMRI currently. There are significant challenges in EEG fMRI, and put it all together will increase the challenges on that.

Together we'll increase the charge of the lot.

Okay, thank you.

Okay, everything's fine. [crosstalk]

Give us a minute while we setup. [crosstalk]

[inaudible] [crosstalk] Shhhhh.

Please respond and we'll see what the [inaudible] typical reaction doing. Aren't we doing [inaudible] We want go below like --

So now introducing Faranak Farzan for the second talk of the day and the last talk and we'll answer questions after. Thank you.

Alright, well thank you everyone. It's a pleasure to be here and it's difficult to go after such an inspiring talk and try to both keep you awake and inspired. So, and on top of that move to EEG, which we know usually loses the fight to MRI.

So, bear with me, I try not to go too much into detail but sort of paint a bigger picture of how [inaudible] could complement some of the ongoing work and what we're doing. I'm currently, now -- I recently moved to take on a new position as a Translational Scientist at Simon Fraser University in West Coast, Canada.

So there, we're doing some interesting work that I mention at the very end of the talk. Bringing several partners together form healthcare authorities to community based research and quite a lot of neuroscience research, including combination of system modalities to sort of bring and address a societal need that we currently have in our part of the country. And that is mainly addiction in youth and also depression in youth.

So, we have quite a bit of young individuals who overdose in the past one year or two and this is in part how we're recruiting scientists around the world to come and work there and sort of try to address that. And combination of different imaging, sort of try to understand how the brain works and how we can offer new treatments is one of the initiatives that the government and the city that I'm currently in have pointed more resources toward. And during this talk I try to wear a lot more of my translational hat and try to sort of paint the picture that we need to come up with technologies that can [inaudible] today be used in a clinic or in a day-to-day sort of basis. Enhance the interest in the EEG and more sort of reportable versions of our imaging technology.

I hope it's not too simple or too boring. I try to again stay general, but feel free to ask questions after and we can get into the details of all that's there to do this kind of work.

So, the context I've painted is the high rate of depression that we're dealing with globally and particularly in youth. In a home depression actually peaks and starts. I was reading somewhere, that actually up to 75% of mental health disorders kick in, in the age range of 15-24 and that's our very vulnerable population. And that's where we're trying to come up with new treatments and diagnostic methods.

And particularly addiction is also something the co-occurs a lot with depression and they have a lot of underlying mechanisms that are shared. So the technologies that we try to develop is sort of added to societal need.

And the problem is obvious, patients come to us a little bit too late. If we were going to sort of come up with this hypothetical measure of brain health that every one individual followed over time, this kind of index or what happens is that individual normally seek help or come for treatment or diagnosis usually when it's too late. And the brain has gone through several changes due to the underlying problem.

So this is obviously one of the problems we deal with and this is why we cannot prevent suicide at the moment. And another part is that the treatments we have is not as effective. Specifically when it comes to the younger population. We're borrowing a lot of the treatment from the adult population. In part, because we're not sort of looking at biologically targeted treatments, per se yet. And we sort of use serendipitous findings to sort of treat some of these conditions. And the numbers are quite high, for youth depression is up to 40%-50% of people are not responding to antidepressants, the medications or psychotherapy. And some actually form suicidal ideation and side effects.

And then for use, it's less available, the option of sending them to seizure therapy, I mean even in the adult population. Only 1% of the population receives this treatment, although it remains to be the most effective treatment. We still don't understand how exactly it works. What kind of network to the brain that changed following this treatment. So the technologies that we developed, we also hope to sort to address this. Can we understand how seizure therapy works in order to offer new treatments today.

Alright so ideally, in an ideal world we would like to sort of understand, which treatments are better for which individuals -- So that area of precision medicine obviously. So, which we are trying to, I'm sure globally we are all trying, to get data into our research, and try to exactly [inaudible] looking at networks in the brain with the [inaudible] MRI state [inaudible] and the status of mind.

So, this is our first step that seems to be going in a good direction and it is all [inaudible] here. And another part is using these neuro markers by their base on EEG or MRI. Where we know things are happening, use them and capitalize them in design of new treatments. So, could it be design of new RF MRI coils or it could be design of [inaudible] protocols so that we can actually target our underlying problem.

So, ideally we'd like to be able to one day use our technology to help over time detect when the problem arises and then fit them with the best [inaudible] protocol or brain stimulation protocol that could sort of moderate that impairment. So, this would be our ideal scenario and our approach, as I mentioned, more of a translational one that is "Can you take the very exciting -- I don't know what that is at the top --

I'll get that out of the way. Sorry.

But I don't see it on the monitor actually. It's fine.

No, not your top. You're middle, the middle thing, recorder.

Oh, this one. Funny, that does not show here.

Yeah, it doesn't show on the-

So the idea is sort of to go into fundamental research and sort of try to translate that into basic science. Sort of research inside the lab and then try to translate it outside the lab and into the real world. So, I'll get into that, how we're trying to form partnerships to make this happen. But, also speak to how -- Now it's not moving anymore --

Think you have to click back on image.

There we go. And obviously the challenge in realizing that vision is the complexity of the brain and it's not just as easy as putting some headband on the head and trying to measure things. It's way more complicated than that. And we need more [inaudible] collaboration across our colleagues and across the world and between disciplines.

So, some of the movements that I'm sure happens across countries is sort of this merger of research between neuroscience and engineering and medicine. When I was at University of Toronto, I was one of the only engineers in the Department of Psychiatry. And now, I'm one of the only scientists that does mental health work in the Department of Engineering at Simon Fraser University. Which is very exciting because you go in and you look at application of all these kinds of different things that then you can use in a different way. It's amazing to be in sort that position actually.

So, in order to make this work we need to work cross sectionally between Technology Development and Treatment Development because when we go into our pillow and try to come up with technologies that may not translate into clinic, then we have a problem. So we're really trying to work with people, subject experience and help out authorities and everybody at the same time to sort of paint this picture.

And going into some of these technologies is obviously, I was in part invited to speak to Kenneth E -- I'll get into that as one of the technologies that I'm redeveloping and then I leave the rest for maybe future talks around some of the more portable options that we haven't worked in the lab.

So if I was going to put some of the major multimodal sort of efforts here on one slide. It's a combination of imaging, whether it's structural or MRI, electrophysiology, neuromodulation, and then the combination of all of them into what we would call a non-invasive multimodal approach in understanding brain dynamics. If we can get into TMS EEG right away as our method of interest today -- One thing unique about TMS EEG is that you can sort of stimulate the brain with some sort of a base form in a way that we used to do that in a Petri dish and in single neurons that are recording in a lab, and really TMS EEG sort of just sort of allows you to do that and it's a very good high [inaudible] solution. In essence, I really see that this is the human version of doing that type of work in today's age. And of course, when you combine it with an MRI, which we do in some of the studies that I mentioned, it then gives you that added advantage. Now not all of them have to happen at the same time, it's possible to get the MRI on a separate day, understand the resting place connectivity and then view that sort of know where to go with your TMS and how to then to coordinate simultaneously.

So we don't necessarily need to get caught up in the technologies of combining everything but this work really translates some of those fundamental researching into human research in wide awake humans. I think this audience probably knows a lot about TMS so I will just get through how the EEG fits into this picture. TMS is stimulating let�s say a motor area and then you can get an equally spinal cord if you have electrodes there.

Well, there has to be some sort of EEG at top but maybe it's not there. But, EEG falls on top of that. Believe me.

So in terms of TMS precision, so that's a good question that was also mentioned - How precise is TMS? So there are chronological studies that look at impact of TMS and pure pulse TMS on different neurotransmitter systems and based on that, several protocols in TMS world have been created to look at activation of GABAergic inhibitory systems or an [inaudible] systems, or even the connectivity between hemispheres. Based on the studies we sort of have an understanding of TMS on average, not in all individuals, but maybe in half of the public research subjects that we test could sort of give us these [inaudible] of GABA A, GABA B and interhemispheric connectivities that may trade stuff -- Health in the motor system.

If we move then into some of the recent work that's been done using TMS and single neuron recording in awake non-human primates, we also see evidence that single pulse TMS seems to reliably activate neurons but it has different impact across different neuronal population. Which is interesting. So as long as we know which neuronal population we are stimulating or which one we want to stimulate, I think that this is promising in a sense that we can stimulate that the same way if we put design actuals around it.

So, to me, TMS seems to be precise enough for now for some of the work that we tried to do, but obviously combining it with, maybe, ECoG or more single neuron recording type approach would help us to understand a little bit more precision what exactly we're doing.

And then you come to our TMS and it was also suggested in a previous talk, so we can now have sort of different protocols of stimulation. So high frequency and low frequency was mentioned, but even the pattern of stimulation could be modified, different networks differently. So we probably have, in this crowd, heard of theta burst stimulation or intermittent theta burst stimulation versus continuous theta burst stimulation, and these stimulation are often applied at different intensities, so this particular stimulation at theta burst, for instance, is often given at 80%. What is probably the idea that it could be the population of neurons that it activates could be dependent on that intensity, but again, speculation as you said and we need more work in a more fundamental type work to, sort of, state that as a fact. But, you know, take that with a grain of salt, but that could be what it is.

And then from the EEG side then it's the same story as MRI. With EEG -- I think of EEG like a genomic type of thing. So you have all these kinds of signals that you can record from so many channels and from so many different layers of connectivity between them that you can sort of come out with different algorithms to sort of make sense of that and then relate that to changing behavior.

And this is what has been happening in the past few years, so there are local measures of EEG that have been looked at in terms of frequency content of the signal that you're recording with EEG or how it responds to -- Or if you have a stimulus that you're presenting to your participant that evoked response -- Which part of the brain responds to that locally, at the local level, and then you can look at the connectivity between different areas. So you stimulate the region - now has different EEG signals across different regions with different frequency are now coupling together to respond to that event.

And then you can also have global responses, kind of more like the network approach in MRI nowadays. The closest that it may get to the resting state functional connectivity type matrix is the idea of micro state. The [inaudible] in bridge, they look at kind of like a data driven approach of coming up with networks but using EEG, taking all recordings across all electrodes at the same time and looking at their dynamics over time and using ICA and PCA type approach to cross their inter networks and then see how they change. And those are some of the works that we're doing in the context of TMS EEG. And then you can also look at the reliability of the signal over time, which is also something of interest in several [inaudible] disorders.

There we go, here it comes. The one -- When you combine the two sort of techniques that's sort of when you get that added advantage. So you can stimulate the [inaudible] precision, so we know a little bit of that and then you can record all these different signals and you can use an IAF algorithm that you choose to sort of capture the impact of TMS.

But in a more basic form of things, if you sort of want to fit it in words, if motor evoked potential falls you actually do get a sort of phase of activation of excitatory neurons followed by activation of fast acting inhibitory neurons and then slow acting inhibitory neurons. Which, nowadays, following the sort of literature that has come out over the past ten years, you sort of have an understanding that, perhaps, some of these [inaudible] forms are actually related to specific neurotransmitter systems. So the early ones related to the NADA sort of circuitry and then the later ones, such as latencies that come in an EEG around 100 millisecond that can be related to, perhaps, activation of more inhibitory neurons. And more and more research supports that, although it's not cut and dry and the brain is more complicated than that. And once you stimulate the brain you also get into the interaction between different areas. So what is really coming out from the literature, if I was going to give my two cents, is that what we are seeing at the first 100 millisecond perhaps is activation of local areas, local interneuronal population, and then once you get past that you're sort of looking at a more complex interaction between different networks and that sort of gets captured into EEG. And, depending on the brain's state, this could also change. If TMS is outside and wake, the signal that you see might actually mean something different than if TMS is outside during sleep. In which case the connectivity between different areas is actually changed.

All in all collectively however, it sort of fits really nicely as a more direct measure of TMS impact compared to what the literature used to do with TMS applied to motor cortex. I'm looking at more able potential, which would not allow a [inaudible] impact on other brain regions, particularly the lateral prefrontal cortex.

So more and more we see literature and evidence coming out sort of using TMS EEG to link different signals and features that we can capture from TMS EEG with genomics or even good behavior, and a nice body of work is building around that, which is quite promising. And also in relation to understanding the impact of different interventions on different transmitter systems, which is the bottom slides that I'm getting into now, how these TMS EEG responses do change with intervention. And I will get into these at the end of the talk where I actually talk about some of the work we've done with TMS EEG around a [inaudible].

Now first, for those who are interested perhaps in the more technical aspect of [inaudible], I'll get into a little bit into the work we've been doing to make EEG [inaudible] compatible, or TMS EEG compatible, however you want to put it. So the marriage of these two sort of technologies hasn't been easy, if a MRI TMS and a MRI EEG. And then there has been a lot of ongoing fights to make them work together over the years. There are some commercial systems that can make this combination easier, but at the end of the day even them, more often than not, fail in sort of giving you the TMS free system.

So what we've done really with our colleagues is getting to developing some sort of a free or open source software for addressing all kinds of different issues that have arrived. From this combination of this technique, some of them are more basic. In early stages it will be heating of the electrodes, or really, like impacting burning the subjects in. More and more will be learned through it, we understood that noise that comes from the EEG proper channel system really is co-found in TMS-EEG world because you get auditory evoked potential. It's a little bit worse than probably what we're seeing in fMRI, because we're recording simultaneously and within the time resolution that you actually do capture the auditory evoked potential. That has been something to address.

On top of that somatosensory - the effect that the TMS over the scalp could have.

So this is a table that I'm sure is in some of the papers that we have to use but these are all kind of artifacts that you as a scientist who want to use this technique should be aware of. Some of them may not come out right away and you may realize that [inaudible] later. Officially the auditory evoked dots it could be also challenged. To address all of things and to make life easier for our fellow scientists who want to use this, we've been working with some of our students and collaborators to create an open source toolbox and sort of trying to streamline the process of TMS-EEG in-office. This is with the idea of one day being able to use TMS-EEG the same way the resting state fMRI, for instance, is shared across institutes, so that we can build that big dataset that can be used to sort of start them out, categorizing response to TMS into groups, and into categories of -- from activation of different brain regions.

So, this is just highlighting and acknowledging the students and post-docs who have worked on this application, and in the middle highlighting some of the artifacts that you could see. Which again, the details don't matter, it's more sort of -- the paper is published, and we have sort of made this available online for free, and I'm happy to discuss that if someone wants to adapt that into their research.

I'm highlighting some of the features that, for instance, the toolbox has. So, if they put it into the box, and sort of walk the researchers through Step 1, and sort of that approach of taking the raw data, and then sort of processing it step by step. And it provides visual interactive GUI in order to tell you how good you're doing, and how clean the signal is becoming. And does it look normal or not. And we claim this sort of shows that you can use any sort of TMS protocols with this app, it could get single-pulse, dead-pulse, normal-pulse, you could sort of use the visual GUI to sort of cut out or decide where your pulse is, and sort of manipulate it or track with it.

And then here, again. So how by removing some of the bad components, you can go from a very noisy signal, slowly, to a signal that looks more like a brain response.

And some of the other features, so you're looking damage analysis, so you understand the difficulty of [inaudible]. And EEG analysis in general is -- there's so many sort of features to look for when you're cleaning the data so we've tried to make it a little bit easier and more intuitive by sort of giving you more mirrors when you're driving a car, so we sort of show you where the really bad files sort of fall, how perky the rest of the trials, and also if you're doing a subject base analysis, it sort of shows you where your subject is relative to the rest of the subjects that you cleaned.

And it sort of makes it a better tool for standardized cleaning of cross-centers, and that's part of the reason we've been coming up with this in order to kind of promote that approach. And also it allows you to sort of maintain more of your signal by virtue of not deleting the entire file. I mean, in EEG world, we have your recording, multiple channels, and we apply, I think, 200 pulses of TMS. So, if one channel is bad, or one trial is bad, we tend to throw out everything. Which we think to be a waste, if you've done TMS region analysis or recording you know how difficult it is. So we're trying to sort of get around that by allowing single-trial and single-channel removal of the bad data, in a way that is still valid and is not sort of [inaudible].

Anyway, technical details. And on the way, as you go from Step 1 to Step 10, you start seeing signals that look more real, and you have some of an automated vision of removing some of the signals, which again, helps with someone who's newer to the field of data cleaning.

So sort of this show, so every column is showing four different TMS protocols, single-pulse, all the way to tera-pulse. And we're showing how different -- following the different steps, you go from Step 1, which basically looks like a TMS artifact, with [inaudible], but large to immicroble scale of signal that looks like brain wave with total graphic plots of change of the TMS evoked potential, which is what you should be seeing in a TMS-EEG study. So all of that is available online, for whoever wants to download it, and you can contact me if there are issues with compatibility. Several people are contacting us. So we're working on that. So be in touch.

In addition to that, we've been also working, so that wasn't a streamlined pipeline that we developed, we also worked quickly with Nigel and Julio from Australia and from Finland, and now I think Washington or Chicago, to sort of create a library of functions that we made available in EEG lab, which is another software for EEG processing, this is more like a library of tools, and if you come up with a new algorithm for TMS-EEG processing, it can be [inaudible] submitted to here, so this is more like a library of tools, and a previous app is more like a car to drive, if you're analyzing signal at the cause site.

All right, so TMS is obviously one of the tools that we've been playing with. It's an idea of using EEG to sort of look at brain health and disease. We've been also developing a statistical framework to sort of allow more for doing these type of work. So one of the challenges we see, when we see, when we do SSI and EEG studies, and being part of several of these committees, sort of enable multi-sign approaches. One of the main issues of the statistical framework that we use in individual studies, which make comparison between the studies almost impossible. So we're sort of coming up with new, unbiased ways of statistical testing. So one way's unbiased, one way's unbiased ways of functional network connectivity, categorization and also I think thresholding is also another thing.

So making ours for that available online material on this, so we recently published something about biased cluster-based estimation approach. We just sort of made the researcher hand it off, around sort of setting a threshold for significance. And it can be applied to EEG or MRI, but we've been sort of dealing with this around multi-dimensional sort of signal that we have in TMS, which involves, not only time, but also frequency, and also space and [inaudible] this object level. So you have this fourth-dimensional data set in which we're trying to apply some sort of thresholding on top of it. And it's really has proven difficult for when it comes to comparing that we've seen of it, students in that lot. So, we're trying to enforce this type of stuff also to make cross-study comparison more meaningful.

And then we're also in parallel working on that, as I mentioned, some portability aspect of -- so the TMS-EEG right now is something is inside a lot. And in my new position we're sort of working a lot with community, also. So and you want to follow individuals in schools, or in addiction recovery programs over time daily, and I cannot take a TMS-EEG to them. So we've been trying to sort of see how we can now see whether our approaches can become portable, so we have students who work for them in [inaudible], and sort of come up with technologies that can capture some of the more established markers into a headset that is wireless-free. It's not easy. So that approach is not very intuitive.

So we've been trying to recapture the newer marker that is reliable with our heavy-duty, high-density system, and we take that and we trying to make that take out as much information as we can until this thing no longer reliably tested over time. And then at point we say "Okay, this is probably the bare minimum we need" and then we go and work for the company, sort of trying to come up with the hard way around it. So, sort of like a [inaudible]. So now let's see how some of these are being actually used, some of these techniques, in the real world, I mean real world research sort of world. So with TMS EEG we've been actually playing around with the idea of [TMS EEG] a lot and trying to put a -- Translate some of the older or more conventional, classical paired policing of [inaudible] and RTMS sort of protocols that were established based on motor cortex. So we've been using TMSH to sort of translate that into the prefrontal cortex and other regions. And to echo the previous talk, a good point that was made is that what we realized also when we compare motor cortex with non motor cortex, we realized that actually these matrix are not the same across brain regions.

And -- Which is a good thing because we knew that as well. It makes sense to invest in these types of efforts, otherwise you can just measure motor cortex and go home because it's so easy to do.

But we do see that once we actually measure these things from one brain region to another, these markers do not correlate. So some object with a very good, let's say, inhibitory marker in the motor cortex may not have a good inhibitory marker in the frontal cortex, so -- Which makes it all more interesting. It's a non-linear thing that happens.

So, maybe instead of trying to look at this complex picture of -- Sort of playing around with different inputs, though, kind of really like, kind of like a [dosed response care] but with different markers of health, so. Inhibition, excitation, plasticity is a big one that we've been trying to sort of look. And then we have a major issue here in the brain states and I think this is maybe what resting state, also should incorporate more into the work and that is how do you make sure that people are in the same state when you're studying them.

And then across different stages of a disease, that's another thing. So trying to capture that and coming up with ways to keep that constant.

And then the output by itself also, the whole neural, how do you capture that output, can be done with even with a [inaudible] but with EEG, you know, we have all those metrics that I mentioned around local response or global response and really this picture is forming by efforts of some scientific [inaudible].

So some examples of some of these markers with the TMS protocols. So this is a study that we've done back in 2013, which is four years ago now, with combining one of the classical TMS protocols. You guys, have you guys heard of protocol [inaudible] before? Probably everybody? No, maybe. Some people say no.

So in this protocol, so what happens is that a supra [inaudible] TMS is applied to the motor cortex. When the participants are sort of engaging -- Activating the target muscle of interest, they could be holding a squeeze ball in their hand or it could be a force meter. But in any case, if you look at the EMG, the electron myograph on the screen, it looks something like that, we see the TMS pulse comes, there is a big [variable] potential but after that, what we see is we see a silence even though there is -- The subjects are still holding that ball, squeezing it hard. We [inaudible] electromyographic activity and then after some time, and this time varies from subject to subject, from state to state, the signal comes back on.

So this period, during which there is a silence of background EMG, has been called cortical silence period, and in the motor cortex TMS world, it has been shown that it varies across different disease conditions and it can vary following from [inaudible] intake or if someone takes some of drug. This can be manipulated.

So, and it stays [inaudible] sort of related to activation of gabaergic system at the central level. There are peripheral contributors to it too but it's been very strongly linked to activation of the inhibitory system, specially the GABA B 00:04:10 activity.

So we asked ourselves, well now, if we measure EEG at the same time as this protocol concurrently, what do we see around this 100 [inaudible], during this silent period? How does the brain look like, what's the state of it, what is the signal that comes from it? So these are topographic, this is like a movie of one individual subject, and there was a little plot star, the topographic plot showing where the maximum signal is at every point in time. And if I zoom in that silent period area, what you see is that during that time, subjects are in this configuration and that configuration is -- Sort of stays stable until, you know, the other configuration.

And that happens at the front, centrally, and peak is around 100 milliseconds. And this is an example of one of the study in which sort of points us to the direction of this signal in EEG TMS and 100 could be used maybe as a proxy, or proxy measures of inhibitory activation.

Now, again, this could be linked to other stuff too. It may not be a pure measure of inhibition but as I've shown in some of our studies, it's a very interesting marker, it has a very good signal to noise ratio and sort of predicts a lot of things that sort of begin to make sense that it could be a measure of inhibitory activation.

So this is one example of one of the protocols that we used concurrently with TMS, EEG and GNE and we've done this work actually over the years with multiple of these protocols. So if your [inaudible] similar with paired [inaudible] protocols, for instance there is one in which I showed a picture earlier in which [inaudible] threshold stimulus are applied, 100 ms apart to one brain area and what you see that, is compared to one pulse alone, the response with two pulses, when you look how motor potential is the suppression of activities. So what happens there is likely that a first pulse activates again inhibitory neurons which springs this inhibitory state so that the second pulse is -- No longer does what it's supposed to do.

And, so again, that could be comparative across participants and across brain regions sort of to see how inhibition works around the brain and across individuals, across this.

So what we did here, we compared that marker in motor cortex with the dorsolateral prefrontal cortex in this state. That was done years ago and we [inaudible] and we sort of showed that indeed there is inhibition in the prefrontal cortex but when you compare individuals between their brain areas, it's not near. So it's not the same. So an individual who has a high inhibition, let's say, if I can call inhibition in motor cortex is not -- Doesn't necessarily have a high index in the frontal cortex.

Which again points to the fact that it's interesting to [inaudible] work and EEG also allows us to look at different frequencies also. So with EEG we have a good measure of looking at now different frequency of oscillations in the brain. We've been looking now at how this pulse protocol could have sort of modulate the oscillations in one brain region. And we've been playing around with that across brain regions.

Sort of similar to the SMRIs that we've been looking at into hemispheric sort of activation of signal. So let's say you stimulate over the left DLPSC. Now what you see over the contralateral DLPSC. So we've been actually combining DTI with TMS SMRI in a non-concurrent way but we've used DTI to sort of understand what is -- Is there a correlation between the integrity of [inaudible] across different parts of corpus callosum and how much signal gets through. And we've actually seen very regional specific changes that. So depending -- So there is indeed cross-activation and this cross-activation seems to associate very nicely with the integrity of corpus callosum.

And it's -- And again there is differentiation between different brain regions. So meaning that you cannot understand what happens in the prefrontal cortex by just looking at the motor cortex. So this study shows that and that's the one that I actually showed you around the cortical silent period.

We've been also doing recently some work with RTMS. So what happens if we do RTMS EEG? So this is a work that was done with collaborators Mark [Alcrow] and Jeremy [Schmalen] in NGH so, in which we sort of wanted to see if we use functional resting state MRI and use a sort of network guided approach and stimulate default motor network versus frontal parietal network in the cerebellum, what happens in the cortical area? Are we going to get activation of these specific networks using EEG?

And the answer is yes, it's very interestingly, we sort of got activation of very selective networks depending on when -- where we stimulated the cerebellum, and it was easily picked up by EEG. Here we're looking at the complexity of the network, which is a marker that we use for EEG analysis. But again more sort of evidence towards the direction of TMS EEG MSRI combination, sort of giving us more clues around the activity of network and its modulation and manipulation by RTMS.

All right, so have these -- So this is fine, this is still in the domain of basic science kind of stuff. So can we now use this in a clinical population in a meaningful way? So in fact TMS EEG and RTMS have begun to be used across [inaudible], so -- Some studies published, some not that you hear from in conferences, obviously. And I think -- I've seen it's actually been coming up in [inaudible] news, and also in all kinds of problems across the neuropsychiatric domain all the way to dementia. So it is being used in order to understand different pathways, and some of the work we've done, we've played around a lot, so I have the mental health sort of research domain. So I work a lot in this domain, and some of the work done has been around the idea of coming up with diagnostic markers for different neuropsychiatric diseases.

So we've been playing around with the idea of impaired modulation of oscillations for schizophrenia, bipolar disorder, OCD, and depression. So it's the work that's been building up over the past ten years in which we look at how, for instance, an inhibitory marker in TMS could modulate this network of what is impacted, and can we come up with ways to differentiate diagnosis and now more with the [inaudible] approach, we're getting into more subdomains of behavior. What is -- What can we use as some sort of TMS-EEG marker of suicide, or anhedonia, or things like that, or inhibitory control.

We have published work in this area, and which, if you're interested we can -- I can refer you to those. And then in children and in youth also, some very interesting work coming up. As I mentioned earlier in the talk, treatment-resistant depression is a major issue, especially in the youth population. Some of my colleagues at Mayo clinic -- I'm not sure if you've heard the name Paul Croarkin. He's been trying to sort of use TMS in the children's population, which you can -- I'm sure you can appreciate how there are so many layers of difficulty when it comes to that age range, but he's done some elegant work showing that actually those who do not respond to pharmacological therapy may have this distinct TMS markers that could be potentially used one day to say, "Okay, so you're not going reply to pharmacological treatment. Let's maybe move to TMS or something else right away." Some of these current pulse measures seem to be impaired in those who do not respond to pharmacological treatment. Again, these studies are small, but we need more multisite trials to build up the numbers and [inaudible].

TMS-EEG has been used in ADHD. Sort of -- keep talking about it -- Inhibition kind of makes sense that that's where research is starting neutralizing it to. Sort of to look at whether it is inhibitory markers that I talked about, or more impaired, or less potentiated, or impaired, or in some ways abnormal in children with ADHD versus normal controls. And these early studies that come out actually point to that direction of inhibitory markers being the ones that differ between younger populations with ADHD and no ADHD.

Now, we've been using these markers to sort of predict. Back to the early slides around context. The prediction of who responds to what is of interest, and also trying to understand how some of these old but sort of more invasive techniques like seizure therapy work. So we've been using TMS-EEG for monitoring and for prediction approach in that context. Design of such studies looks like this, obviously. So, a little bit like what was described in the previous talk. So you do baseline assessment using EEG and TMS-EEG, and oftentimes you actually have MRI as well. In some of these protocols, it can be looked at after for verification of [inaudible], and then you provide intervention and you do a follow-up. And if you have a lot of money, we can do more assessments in between, and if we don't, we sort of do pre- and post- and follow-up.

We're in the middle of doing a lot of this work, and I mentioned some of the published ones or some of the ones that are closest to publication. So this one is the published one in which we used TMS-EEG to look at who responds to this new form of seizure therapy that's called magnetic seizure therapy, which a lot of you might not heard of. And we -- In a study that was done by one of our graduate students, Ying-Ming Sun, it was -- We actually showed that an inhibitory marker that I showed you N100 -- At baseline, the magnitude of that predicts who's going to be suicide-free following magnetic seizure therapy. So, those who had a good N100 in their prefrontal cortex seem to benefit from seizure therapy the most.

Now, why could that be? I mean, it's possible that this marker is sort of a proxy to brain health in general. It means your brain is in a healthier form, so you benefit from this treatment. Or it could be that maybe this marker is a proxy to positioning or some sort of an activation of inhibitory neurons. And the treatment that we have sort of targets that population, and if those populations are accessible as a measure for me why this treatment works the best. So a speculation can go on, but the fact of the matter is that if you're in a click and you want to predict who's going to benefit from this treatment if they're suicidal, we kind of have a translational, or potential translational marker that could be used. And the predictive value was quite high. It was around 90% of accuracy in predicting, which is quite high for the EEG-TMS world.

In another non-published work, in which we looked at the connectivity of dorsolateral prefrontal cortex with the rest of the brain before and after seizure therapy in, I think, 13 or 14 of individuals in this picture, what presented a year ago, I think, at ACMP. So, I wanted to show the movie quickly, and I'm getting closer to the end of the talk, and in which we're sort of showing how the DLPFC is connected to the rest of the brain before and after seizure therapy, sort of following the movie here. So if you look at pre-, you see a lot of connectivity between the DLPFC and several regions proximal or distal to the site of stimulation, which after the seizure therapy you don't see that.

So there has been this ongoing hypothesis that depression is linked, in part, to hyperconnectivity of different networks and interaction between networks, and it's possible that what we're seeing with TMS-EEG is sort of what I call the brain at ease or a calmer brain. So the connectivity is sort of dampened, maybe by seizure therapy, and that's what we're seeing here. Anyways -- Sort of a demonstration of how TMS-EEG in a limited millisecond time resolution could be used to look at connectivity in the brain following an intervention.

This is another work that we recently published, and we're looking here at not TMS but EEG alone in the context of seizure therapy again, and we sort of coming up -- So this falls under the category of some of the translational work we're trying to do so that we can actually build some sort of a device that inside of -- inside a connect could be used to predict who's going to respond to treatment and who's not going to respond to treatment. Here is a little bit different approach in which we used variability of EEG as a marker to understand why seizure therapy is so effective, and we looked at variability of EEG in relation to change in depression scales and also change in cognition scales.

For some of you who may not be in the psychiatry field, seizure therapy remains one of the most effective treatments, specifically seizure -- ECT or electroconvulsive therapy, but it has a lot of cognitive side effects. But here in this work we actually showed a marker which is an entropy-based marker in EEG which predicts, actually in a regional-specific and in a very timescale-specific way what is the impact of seizure therapy in relation to change in mood and in relation to change in cognition. And these markers are -- The predictive value are still quite high as well, and we're hoping to use this in a context of developing better seizure therapies that are creating less cognitive side effects but they have maximal antidepressant impact.

Not getting into details too much, and as I mentioned, we're building all this streamlined and standardized approach to fit it into our international and national collaboration, so I'm leaving the electrophysiological efforts of CAN-BIND which is a Canadian biomarker integration network in Canada. We have several clinical trials as part of this work, so not only using antidepressants in depression, but also cognitive behavioral therapy, cognitive remediation, rTMS seizure therapy, ketamine, and all that. It's funding that comes through our province Ontario, but it also funds multisite stuff. So what we're doing, researching in a very standardized way, we are measuring EEG and in some of the work, TMS-EEG before, during, and after these treatments.

And we're now beginning to sort of look at now, what's the pattern that emerges across different modalities of treatment, and it's very -- also exciting part -- sort of project to be a part of. fMRI is something that we measure in them as well, which can be sort of hindsight and after the fact of combined with the TMS-EEG and EEG measures. All right? So very briefly, I mentioned youth in my studies, so in the early talks, I'm not sure how much time we have. If you have five minutes I could quickly also show how your -- do we have five minutes?

No one's objecting; no one's leaving. I'd continue.

So we've been sort of extending the utility of rTMS to the younger population in one of the trials we recently finished, an open label trial. The safety of rTMS in youth has been very recently and sporadically established across the world sites globally, but the data is very thin, for obvious reasons. Occasional brain stimulation in kids is frowned upon, but there are a number of them are not responding to treatment so it's an area that we are moving into. And it's a team of scientists that I worked with closely at Center for Addiction and Mental Health, University of Toronto, Mayo Clinic in order to sort of run the first short theta burst stimulation rTMS in youth with depression. So it's a fantastic team of people, just acknowledging them, who helped with this.

So the thing about the trial we did, in 20 years, age of 16 to 21 or 22 that we presenting to a two week trial of receiving theta burst stimulation to see if -- And these are individuals who did not respond to anything, so they were desperate for some sort of treatment just to see what their options were. Given my neuroscientists or translational scientist [inaudible], I put the poor things through so many neuro-imaging, bio markers, some type of work up to understand how the treatment working. So the design looked something like that.

We had 20 individuals going through rTMS theta burst stimulation treatment. At base line, we measured several genetic factors, several imaging, resting state functional MRI, [inaudible] then they would go through two weeks -- Throughout we would do a EEG sort of resting state EEG continuously, just to see what is changing in the brain over time. Then we had some comprehensive post-treatment measures.

The design of the study, for those interested, theta burst stimulation was applied bilaterally; one form was applied on one side, another on the other, which I can get into later. They were also doing some sort of inhibitory control top when they were inside the scanner, given the idea that this rumination that we've seen in depression could be linked to inhibitory control problems, problems with impulse control - which TMS seems to actually target, based on some of the studies that have been done before.

The final results, surprisingly, even though those are individuals who do not respond to anything and they're sort of on their way to either be -- Even seizure therapy or something else -- This is just two weeks ago, and usually in TMS world we do four weeks of rTMS therapy. Chronically, a lot of them actually started improving even at week two, even at first week sort of stuff, so this Y axis is changing symptoms and X axis is the time and what we see here is that quite a few of them actually got very close to remission following only two weeks of treatment. And four or five of them actually remitted. So this is the feasibility stay again to sort of understand the safety of applying this in a 16 year old, but these results are actually quite interesting. The other measure on the other side is showing the children rating for depression, so just to compare with [inaudible]

So we looked at TMS/EEG markers and interestingly enough, they followed our hypothesis whether this is lucky or there's some truth in here. So what we're seeing is that comparing, so this is again looking at -- I'm highlighting again that inhibitory markers because they stayed consistent across the top -- That inhibitory marker we have thought that based on studies that have been done before with TMS, motor cortex in youth, [inaudible] in depression that was shown to be impaired, and this is, in fact, -- What? They want me to go home. -- Is actually what we see, we see that the amperture of that N100 response is actually smaller in this youth who are not responding to other treatment and that following -- So we're comparing them to healthy controls of their age and following treatment, it seems to be corrected for.

The numbers are small and I hate to draw conclusions here based on small numbers, but In the TMS/EEG world, we are kind of used to these sort of numbers and it definitely begs for further expanding the sample size and getting [inaudible] on controlled trial sort of understand if these markers stand, and whether in fact this marker can be used as a proxy for treatment of depression or youth depression in general following studying the other population as well.

Here we looked at left DLPFC and right DLPFC, again, for obvious reasons, because things do travel, and it's good to know. And in depression, we've seen impairments of both sides or some sort of imbalance between the left and right, so and our markers are more specific for one brain region than the other and sort of markers specific for a different brain region.

Same applies when you looked at no-go inhibitory control top without TMS. It sort of follows the same general rule of inhibitory control being a [inaudible] issue in this compilation. We sort of -- Fits nicely with what we see with TMS/EEG [inaudible].

Anyways, so to summarize, so this is basically saying it's ongoing. We are bringing all the pieces together, analyzing all resting state of MRI to see if whether we can use that to say something about the networks that are impacted, such as [inaudible]. So, hopefully you'll hear from this soon in some sort of published form.

So, in summary, this is our vision, and where we are right now. We're trying to come up with some sort of translational way to make it happen, to be able to monitor people over time. Hence our interest in EEG in some ways, given its non-[inaudible] approach. And we're trying to realize a dream of one day having that cardiac monitor inside our family clinics, which we don't see, and it makes sense to have one there, so that's where what our bigger research program is thriving to.

And TMS/EEG is one such approach. It's not portable, but it can be used to come up with portable markers, given its EEG component, and some interesting studies are coming out and, in order to do that, we normally need collaboration between [inaudible], but what we realize also we need collaboration between different stakeholders.

So where I am right now, as of January, that I moved to British Columbia, we have a very nice collaboration, actually, between the government and the city of Surrey, which is where I am right now. It's next to Vancouver, it's like 20 minute drive. So the city has realized the issue in addiction and youth mental health in general and they put together partners and created scientific positions and helped also with establishing imaging centers inside the university, so right now we actually have, we have now is brand new MRI, MEG, EEG. It's very nice big data cluster for data analysis, one of the biggest in Canada our university received. And hopefully soon, TMS/MRI, we would need to recruit scientists to do that type of work, but inside the university we're still establishing that core, sort of communicate then with our other partners. I have an embedded lab at this addiction recovery program right now called John Volken Academy.

It's a two-year addiction recovery program and I'm working very closely with the clientele to sort of -- And now youth are sort of portable or translational markers to monitor health over time in these individuals, and for which I'm going around and partnering with different scientists to do different types of technology development to bring it to that unit, and it's a very exciting work to actually work with individuals who are living with this experience, and who will tell us where to develop technologies and where not to because there's no way they're going to use it if they weren't end users.

And on top of that, we have right now what's called West Coast Health Accelerator coming to our district and it's part of what we call Surrey Innovation Boulevard. So we have companies opening shops there. Health-tech companies. From what we understand, it's projected that up to 40% of companies in Silicon Valley are going to be hub-related or something. Word of thumb, but yeah, don't quote me. So we're preparing for that era, so our students right now are being, in engineering, they're being recruited to Tesla and whatnot, and the future, we think they're going to be recruited to some sort of health tech, brain monitoring, brain health advancement company like Neuralink that Elon Musk is talking about.

So we're building the capacity to bring all the partnerships together and if you guys happen to be in Vancouver for the human brain mapping, I'm happy to show you around and show you our new states there and where we're doing all this work and the TMS/EEG lab that we just purchased the equipment to get up and running as well. So, thank you, [inaudible], the end. And acknowledging all their partners, so I've traveled a lot over the past ten years from universities, to institutes, so a lot of people to thank that I work with, and a lot of funding agencies as well, and even some people in this room that I work with over the years. So thanks for listening to me go on and on and on, and I hope I didn't bore you, but feel free to call and email me or let me know if there's anything that interests you [inaudible]. Thank you.

So, unless there's not a lot of questions, then just please type into the chat window. It's the easiest way, and if anyone else has questions, please --

I wanted to ask-

Go ahead.

So for the sake of portability have you looked into direct current stimulation where you can use the EEG electrodes to deliver stimulation? That eliminates the whole TMS treatment from the equation.

Yeah, I've seen it and I'm a user of TMS more than DCS over the years for the obvious reason of TMS actually stimulates the neurons and transcranial electrical stimulation only modulates the membrane threshold to a certain extent, right? And we are playing around with that, but the robust -- The reliability of that approach has been something that made me stay back and be more on the observation side at the moment.

What we're hoping to do is one day develop something like portable TMS. That requires working with quite a bit of our engineers in the university to sort of [inaudible] power the electronic problems of portable TMS. I mean, obviously, it's three tests lab, one and a half tests lab, jolts up [inaudible] that you need to deliver.

So that is that, but I do work with that technique as well, and I think it has quite a bit of value when you're dealing with populations that, at least from the clinical side, are not as much impacted, so when you're dealing with someone with tremors and depression, for instance, who's on the path to get seizure therapy, kind of need a stronger sort of stimulator to really stimulate, but I think that is a good direction to go [inaudible] what happens to tDCS.

In the context of tDCS, combine it a lot with cognitive training, and what not -- So that's another thing I didn't cover, but we're also developing cognitive training which sort of, in a backward way, trying to entertain certain circuitries and networks with behavioral cognitive training, and then combine that with directed stimulation of that network, be it better with tDCS or --

I just wanted to get your current thinking on the use of auditory masking. What are you-

So what we use right now, so -- Auditory masking has pluses and minuses. We built the capacity to use auditory masking but our experiments were very lengthy in terms of protocols, and we got a lot of complaints from our subjects that they really cannot stand the auditory [inaudible] masking.

So what we do is we apply a lot of sham conditions and then we use some type of foam between the coil and the scalp in order to minimize the bone-conducted part of auditory masking. Now, in studies that are shorter in duration, I highly recommend [inaudible] of auditory masking, but for others, in richer sort of, you know, have an intervention, and you have to stimulate the subject for 5 hours, the auditory masking can become very tiring and I would argue even change the brain state to some extent, too, so.

You can pair that up with sham stimulation a lot of times, even though it's not perfect, but -- [inaudible]

One of the questions regarding the stimulation amplitude estimation. Could you maybe use TMS/EEG to sort of find out what the minium amplitude that they need to use in TMS in order to get an effect in the brain?

Yeah, I think there the question is, "Can we do this closed in TMS/EEG?" So there's some [inaudible] if you know of him from Germany, who did actually doing the TMS/EEG in closed system and there's a company that offers a much easier partner around it. I think the problem with that is, that in order to do that, you need to do, at the moment, impact if you were going to do it today, you kinda have to do the TMS/EEG one day, then look at the data, and then bring the subject back and try to work with it that way. So it's very tedious in terms of design, and I think to get around that you sort of need that combined TMS/EEG approach and there you get into a problem of noise also, right?

So EEG is prone to noise. So you need a lot of trials in order to get a reliable signal. And because of that, you know, the experimenter links here, and so you kind of need to, if that's okay with you, to [inaudible] do that. So hindsight we've looked at some of those things, so we stimulate, let's say -- We can also use something else. We can also use maximum current that you induce in the brain area and use that in a sort of intensity of TMS, to sort of, homogenize or understand how much you're stimulating the brain region. That's another way to do it, but yeah, amplitude of TMS. [inaudible]


Okay. Hope to see you in Vancouver. Thank you. End of June. Well, very soon.