Welcome. My name is Paul Vespa. I'm the Assistant Dean of Research in Critical Care at the David Geffen School of Medicine at UCLA and it's my pleasure to introduce our speakers for today's topic session called Brain Monitoring in the ICU of the Future. Our first speaker will be Dr. Jan Klaassen. Dr. Klaassen is Associate Professor of Neurology and Director of Neurocritical Care at Columbia University and his topic will be EEG to detect and treat seizures. Our second speaker is Dr. Dai Wai Olson. Dr. Olson is Professor of Neurology and Neurotherapeutics and Professor of Neurosurgery at the University of Texas Southwestern University and his topic will be Automated and Quantitative EEG for Non-Experts. And our third speaker will be Dr. Arti Sarwal. Dr. Sarwal is Associate Professor of Neurology and Division Chief of Neurocritical Care and Medical Director of the Neurocritical Care Unit at Wake Forest School of Medicine. And Dr. Sarwal's topic will be Neurologic Ultrasound in Future Coma. I hope you enjoy this session. Hello everybody. My name is Jan Klaassen. I'm an Associate Professor of Neurology at Columbia University, New York. I want to thank the Society for inviting me to talk on EEG to detect and treat seizures. Here you can see my disclosures, my funding, and I'm a minority shareholder at Ice Neurosystems, but these don't directly apply to this talk here today. EEG is quite an old technique. It was first invented in the 1920s, but it was sort of, I want to say, stuck in paper because initially it was recorded always on paper and that limited its applicability, particularly in the acute care setting. With the digital revolution in the 1970s, it was increasingly introduced to the ICU context. What it allowed us to do was to apply advanced filtering, even post hoc filtering, changing the montage, and doing quantitative EEG analysis. I will talk here today initially about seizures. I will talk a bit about depth seizures and the integration into multimodality monitoring and finish by talking briefly about the intraictal continuum. I'll start out with showing you a case. This was a 57-year-old gentleman with VT arrest. He had a rest time of 15 minutes, underwent hypothermia treatment or temperature modulation. He was cast and transiently needed an IVP. When he was rewarmed, the EEG changed from this pattern to this pattern. This is non-convulsive status epilepticus. The patient did not shake, but had this highly epileptiform pattern. It was controlled with a midazolam drip, valproic acid, and levotriazetam. Here is the MRI that we obtained. There was relatively little changes from hypoxia, so attributable to hypoxia. The NSE peaked at 24. He had pupillary reflexes, no corneal reflexes, and no motor response to painful stimulation. When we started weaning the midazolam drip, this even uglier EEG was seen, and so he returned into non-convulsive status epilepticus. Many people would stop there, but I want to encourage you to not give up too early on these patients because as we controlled, again with a midazolam drip that later on was weaned off and adding lacosamide to the anti-epileptic regimen, this was the EEG we obtained then, and he was discharged to rehabilitation and had some mild cognitive deficit as an outcome. Seizures in critical care are common. Most of them are electrographically only. They are not associated with shaking. This phenomenon was initially described in patients treated with benzodiazepines for convulsive status epilepticus, and DiLorenzo found that almost half of the patients had ongoing non-convulsive seizures. In the neuro ICU, medical ICU, and surgical ICU setting, a number of investigators showed that we could frequently see patients with electrographic seizures that were not associated with shaking. If you look at the frequency breakdown, you can see that these electrographic seizures can be seen, including non-convulsive status epilepticus, in a number of acute brain injuries, but also beyond acute brain injuries. You can see this in particularly, for example, patients with cardiac arrest. So if you start monitoring these patients with EEG, you will ask yourself, how long do I have to monitor until I get the first seizure? Pandian, many years ago, looked at this and found that if you only used routine EEG, so 30-minute EEG, you would miss a number of these electrographic seizures. We found that you had to monitor at least up to 48 hours to detect a significant number of the seizures that you would ultimately find. And then the Boston Group did a very interesting study and they looked at the first 30 minutes of EEG recorded and found that the yield of ongoing EEG monitoring, even if you didn't see seizures in the first 30 minutes, but it mattered whether you saw epileptiform activity in the first 30 minutes. So that may allow you to target patients that are higher yield for ongoing long-term EEG monitoring. When you generate these large amounts of EEG data, it is very labor-intensive for physicians and to look through all of the EEG, one way to speed that up is to use quantitative EEG analysis. So there's a number of ways that the EEG can be analyzed. Here you can see, for example, in the middle of the panel here, you can see that this is a compressed spectral array where the EEG is broken down into different frequency waveforms. So here you have slower frequencies at the bottom of the panel, higher frequencies up here, and you can see the color indicates the power in that frequency range. You can see these seizures that occur very frequently here in this patient very easily. You can also display that in other ways. This is just the way the maximum amplitude or the majority of the amplitude is. And here you look at the arrhythmicity of the EEG, and you can see very easily that in distinct frequency spectra, the arrhythmicity increases. So these approaches allow you to quickly screen large amounts of EEG data. Here's another example shown with a patient that had a right MCA stroke and had very frequent seizures, and then when the midazolam drip was started and the dose was increased, you see how the seizures initially become less frequent and then are controlled. There's a number of ways that you can demonstrate this. Here's a patient that had these frequent electrographic seizures at the top of the screen. You can easily depict them even on this quantitative EEG panel. Here's a patient that had frequent electrographic seizures, then had respiratory acidosis because of decreased breathing from the frequent seizures, which suppresses the seizure activity after the patient gets intubated, and that gets corrected. The seizures and status recur, and here's where the midazolam drip was started. Another example is shown here of a patient that underwent emergency c-section for eclampsia with very, very frequent electrographic seizures, and only after the c-section was performed the patient stopped seizing. So these allow you to rapidly screen large amounts of EEG data. All seizures, these electrographic seizures associated with worse outcome, there's a number of pathophysiological associations that have been shown. In ICH patients, edema and increase in midline shift have been associated with electrographic seizures. In TBI patients, metabolic crisis, and apocalyptic atrophy in this study by Paul Vespa has been nicely demonstrated where patients that had electrographic seizures on the same side of the seizures had atrophy on follow-up MRI studies, so there seems to be structural changes associated with these electrographic seizures. Similar phenomena have been shown in subarachnoid hemorrhage patients and subarachnoid refractory status epileptic patients. In patients, in pediatric patient populations and adult populations, electrographic seizures have been increased with a higher likelihood of worse functional outcome. What is the underlying link between all of these brain etiologies causing electrographic seizures? Patients that have, so these etiologies are all associated with brain inflammation and patients that have more inflammation seem to have a higher degree of these seizures. This is being investigated now in the epibios trial and in traumatic brain injury patients amongst other things as a cause for electrographic seizures. How do we treat electrographic seizures? First we'll treat electrographic seizures similar to convulsive seizures and status epilepticus. First line treatment is lorazepam IV or midazolam IM based on these large multicenter trials. Second line now with the recent publication by Kapur is either phosphonatone, biproic acid or levotracetam, they all perform similarly. Refractory status, so seizures that are ongoing despite these initial treatments are treated typically with continuous IV midazolam or continuous IV propofol drips and super refractory status or recurrence of seizures after treating with these initial anesthetic drips are treated again with these drips or also with continuous IV barbiturate drips, thiopental or pentobarbital drips. What I found personally very useful is when weaning from these drips, starting phenobarbital may make it more likely that people don't have recurrence of these electrographic seizures. Just as sort of a tidbit, earlier initiation and higher dose of anti-epileptics early on may be promising approaches, there are no randomized trials but in refractory status and refractory seizures this seems to potentially be beneficial. And considering ketamine because this has a different mechanism of action where it allows benzodiazepine receptors, the GABA-A receptors to become more responsive to benzodiazepines again and not having the side effect of causing more hypotension and if anything causing the opposite and hypertension makes this an attractive choice also as add-on therapy in these patients with refractory status epilepticus. A couple of words about invasive brain monitoring, so invasive brain monitoring typically includes a number of physiologic parameters that are monitored, you can measure pressure, oxygen in the brain with a number of different approaches and blood flow assessments as well as micro dialysis measurements where you can measure brain chemicals like glucose, lactate, pyruvate, glycerol, glutamate in the brain tissue and several investigators have added this intracranial monitoring probe together with these monitoring parameters. One thing that you can see is that you can see patients that have on the surface not a lot of seizure activity but on this intracortical monitoring probe place you can see these electrographic seizures here. What is interesting is if you look at these electrographic seizures here you can see that they may be associated with a drop in the brain oxygen, this is a drop of the brain oxygen from the mid-20s to 4 which is in most studies associated with causing permanent brain damage. So this allows you to correlate what's happening in the brain at the same time when patients have electrographic seizures. When aggregating a number of these electrographic seizures and focal seizures on the depth electrode Paul Vespa together with us pooled some patients with traumatic brain injury and here we pooled patients with subarachnoid hemorrhage and what we found was that patients with these electrographic seizures even if they were focal had an increase in metabolic crisis and patients also had an increase in the mean arterial pressure, increase in the respiratory rate, heart rate. You see that the ICP slightly increases, that there's a brief increase in brain metabolism indicated here by the drop in the jugular bulb oxygenation, so that's the oxygen measured in the jugular vein when the blood coming out of the brain and a very transient drop of the brain oxygen. What was interesting is that normally in patients with seizures in epilepsy there's a quick increase in blood flow. Here you can see a delayed increase in the blood flow. How are these depth seizures associated with other things? So you can see that a lot of the patients have only depth seizures and patients with depth seizures seem to have a worse functional outcome. These are patients with subarachnoid hemorrhage that we looked at in three months after the injury and we found that patients with depth seizures had a worse outcome, higher likelihood of worse outcome than patients that had surface seizures alone. Best was not to have any seizures of course and severe attenuation of the background was the worst, but these focal seizures seem to be associated with worse outcome as well. I'll briefly talk about the X-interactive continuum. It's a large group of different EEG patterns. They are repetitive EEG patterns, so here you can see periodic epileptiform discharges. They are seen here at a very low frequency, so these are occurring every other second from one to another is here one second. You can see that every other second there's one of these occurring, whereas down here there are at least two and a half of them occurring per second. And what we found was when we associated these patterns with brain monitoring parameters that yes the blood flow seemed to increase with higher frequency discharges, so higher frequency discharges like down here compared to those low frequency discharges, but what was interesting was that at some point initially the brain oxygen was maintained, but at some point it dropped. And so it seems that the increased supply of nutrients and other brain chemicals, nutrients and oxygen that the brain needs to maintain function is outpaced at some point with these very high frequency discharges. Here you can see in grouped analysis that this happens when at the time that these frequency shifts occur, that these patterns become from slow frequency to high frequency, there's a drop of the brain oxygen associated with it. How do we manage the interactive continuum? Much of that is actually still in the research area and is not finally decided. Most people would consider high frequency discharges to be more harmful and treat them more similar to electrographic seizures, so if they are like three per second frequency discharges, those would be treated more aggressively. Others, including us, we try to see whether the patterns, these extra-interactive patterns behave similar to seizures. So one way to do that is you can do a benzodiazepine test, give small incremental doses of benzodiazepine or you can look at the imaging or the multimodality monitoring correlate of these high frequency discharges of this extra-interactive continuum and if they behave similar to a seizure, we tend to treat them more aggressively. Also consider the underlying etiology when making these management decisions. What are the limitations of EG monitoring in general? There's some uncertainty of EG findings, as I just outlined. There's poor spatial resolution, so if you find changes on the EG, it does not necessarily give you any insights about where in the brain that is occurring. You can make some generalized assessments, but it doesn't allow you to pinpoint exactly where the changes are happening. To read all of the EG, as I told you, requires quite a bit of physician resources because all of the EG has to be reviewed and has to be carefully looked at. Some of the new techniques like quantitative EG analysis may allow you to make this more efficient and speed up the reading. There are quite a lot of tech resources involved, so to have good quality recordings, the EG technicians have to make sure that the electrodes have good contact with the brain and especially in ICU patients, which may leave for a CAT scan or things like that or are turned to prevent decubitide. This is actually not an easy feat, so the EG electrodes often become disconnected and then have to be replaced. There's a number of electrode developments, there are some electrodes like needle electrodes which have been developed, which have advantages and disadvantages. For example, if they get disconnected, they may be in the bed. So there are advantages of them. The easier application and reapplication have to be weighed against risks, but there's quite a bit of development in this field. In conclusion, EG is a promising tool for managing ICU patients. It may allow you to detect seizures that you otherwise would not be able to see. Remember, the vast majority of seizures in the critical care setting are electrographic and not detectable unless you have EG monitoring on. Patients with refractory status epilepticus or even more, super refractory status epilepticus cannot be adequately managed without EG monitoring because all of their seizures are electrographic only. So in those patients, you will not be able to manage them adequately unless you have EG monitoring available. The patients, especially when weaning the midazolam or propofol or pendobarbital or thiopental, as I showed you initially in the case, not infrequently electrographic seizures occur and unless you have the patient on EG monitoring, when you're weaning the anti-epileptic medications, the anesthetic drips, you're not going to be able to detect them. There's many questions that aren't answered. The echocontinuum, how do we best manage that? How do we best treat non-convulsive seizures, non-convulsive status epilepticus compared to convulsive seizures, convulsive status epilepticus? Actually there's no big trial that has looked at refractory status epilepticus. There are some that have been published, but they were underpowered to show differences in outcome. And then the role of EG monitoring in multimodality monitoring needs to be determined and needs to be established. We are seeing a number of patterns there that we don't fully understand and in addition to what I've shown you there is also a phenomenon called cortical spreading depression, which may be recorded with certain settings of the EG recording and that may be associated with ongoing brain injury and metabolic collapse. So there is a lot of research going on in this area and the significance and what this means management-wise and prognostic-wise remains to be shown. This is just finally a picture of my team, there's a number of research assistants, research coordinators and my collaborators in this picture and I want to thank you for your attention and hope that this was helpful. Thank you very much. Hi, my name is Dai Wai and I was asked to talk about automated and quantitative EEG for the non-experts. These are my disclosures. I think the only one that's really possibly relevant is Sarah Bell and I have done some research with Sarah Bell. If you like to argue, I'm going to give you help right from the get-go. I'm going to tell you the three things that I'm going to talk about are that non-experts reading EEG is simply a natural evolution of the science. Second, I am a non-expert, but I'm a non-expert in EEG. That doesn't mean I'm an idiot and non-experts generally are not idiots. So this is something that's in our lane. And then finally, I want to talk about the fact that I think this is all going to help the EEG experts achieve the goals they have wanted to achieve for years and years. So let's get started. Imagine that if you think EEG is only for experts, what if instead today I was giving you the presentation on automated and quantitative EKG? You'd laugh, right? Willem Einthoven taught us how to say the alphabet PQRST and now we can automate it. We've got automated and quantitative that can pick up a P2P or a PR or a QRS that can differentiate between atrial fib, atrial flutter and ventricular tachycardia with a high degree of accuracy. We've also seen, in terms of oscilloscopic waveform readings, automated and quantitative interpretation of waveforms like you're seeing on the screen. It can pick up the dichrotic notch, it can tell us the rate, it can tell us the pressure, the systolic, diastolic. So this is not really hard and in fact, technology has advanced to the point that I could ask my smart device how old George Washington was on July 4th, 1776 and within moments I get that response. Think about the technology required for that. Not only does the device have to recognize my voice, it has to interpret, am I asking a question, giving a command? And then when I say July 4th, 1776, those are just a random group of numbers and my accent is slightly different from your accent. And my pattern of speech. So technology has moved us forward. Think about where EEG was when I started my practice. This is the type of device I saw. My job as a nurse was to untangle the stylettes, ink the stylettes. If somebody would come in, the paper always got bundled out. Changing the paper was itself a process. Most of my time as a non-expert in EEG was originally spent babysitting the machine. So I didn't have time then to do some of this others. Over the many years of EEG, technology has evolved. So that's really my first point is we've had a natural evolution of technology. This stuff is moving forward. This train is on the track. Non-experts. Who are non-experts? Well, I mentioned that I'm a neuro-nurse. I'm a non-expert. APPs, neurologists, neurosurgeons, intensivists. All of these people are non-experts in EEG, but all of them have also obtained a college degree and specialty certification. That simply is to support my thesis that these people are educable. They're smart. They just are not experts in EEG. Here's a quick quiz. You have two patients. I give you their screens. I tell you one of these patients starts to have a seizure after three seconds and the other one has isoelectric. Without any knowledge in advance, I bet you could pick out which is which. And yes, I gave you an easy one to read, but that's what non-experts are looking for. Our non-experts are not trying to become epileptologists. We're trying to get past that main bit. Is my patient seizing or are they not seizing? It's not that we're non-experts. We're not experts in EEG. We are experts in other fields. That's my second big point. We're not idiots. We're experts. We're just not EEG experts. Quite frankly, most of us don't want to be EEG experts. That's why we're doing what we're doing. Let's go a little bit deeper here. This is the harder part to process. When we talk about automated and quantitative, automated is an adjective that's used to describe a process that is performed by equipment. So automated EEG is EEG analysis performed by equipment. Easy enough. Quantitative is also an adjective. It's used to describe something that is measured. Therefore quantitative EEG analysis is an analysis that's performed by measuring something. Just to reiterate, our goal is not that we are going to, come 2021, replace everything about the epileptologist. Our goal is not that we can entirely automate thinking and replace critical thinking. This is not our goal. Our goal is probably more like this. We're looking for something that's pretty much an early warning system. I want to know, man, this is odd. I'm picking up something that needs to be attended to. If it's simple enough, I can figure it out. If not, I know that it's beyond my scope and now I'm ready to call someone else. So separate out that we're not trying to completely replace everybody, we're trying to supplement them. We are an adjunct. Here's the basic requirements to understand signal process EEG. If you can get past this, then I think you are educable and you can figure it out. So here's one panel and I just picked that. Over to the right side, you see two different panels separated by a little gray line there. If you can see that one of those is somehow different than the other one of those, then you can probably handle processed EEG content. Here's another image straight from there as well. One part is different from the other part. Now right now we're not going to dig down deep into what exactly is different or how it's different, but I'm just trying to show you that this is what we're looking for on the scope is what's the different stuff. When you think about processed EEG or automated EEG, this is a way of teasing out four parts of a signal, symmetry, left compared to right, amplitude, how high is the signal, how much up and down, frequency, how often, and rhythmicity, how regular. we can look at that with things like 4A transform. And more and more commonly what we're seeing is that it's a combination of evaluating symmetry, amplitude, frequency, and rhythmicity. What we're seeing trend into clinical practice is that instead of just looking at symmetry or just looking at frequency, we see these combinations and we see them presented in a panel of results. These are three different types of result panels that you might see. I will tell you that the verdict is still out on the best visual display, the best graphic user interface that we're going to see. So these are evolving a little bit and they're being teased out. We're kind of at the early stage. But you can see that there's some similarity. It seems like we've settled on certain colors and certain patterns. And we're just presenting these and at some point in time probably one of them will bubble to the top as the favorite style. Sonification. I want to close out talking a little bit about sonification. And today while I'm taping, this happens to be December 22nd, last night everybody went outside and looked up at the stars to see the Christmas star, which is really cool. But the really awesome thing that happened in 2020 was that NASA released the Chandra sonification project. In other words, NASA took their technology, they got an electrical signal from the heavens and they changed it to a sound. This is really cool. And it's also really high tech. Now if you look down in the bottom right hand corner, you can see I give you a link to listen to this and I encourage you to do so. But really if you just Google sonification or Chandra project or NASA sound, you'll get these and you'll happily spend an hour or so geeking out. Now if it's possible for us to sonify the entire universe, it's not a big stretch for us to sonify a little bit of EEG signal. And that's one of the things that's happened is instead of just the mode, the original mode of looking at EEG was a big, huge, long ream of paper with lots of squiggles where the ink kind of came in or came out. Then we went to electronic and we went to more continuous. We're adding these modes of evaluation. Imagine that there was one mode where they could sonify that there was no seizure present. The brain was very organized, like a nice quartet. You could hear that very smooth, very organized, not too bothersome. Imagine the other mode was something else. Could you hear the difference between the Emily Carr quartet and between Rammstein? I'm thinking you could. And all they've had to do is sonify this. This is done because we long ago recognized that time is brain and we recognized that brain is neurons and glia and some other stuff. So from that, we realized that in the critical care setting, time is neurons and when someone has a seizure, it's all about saving those neurons. In critical care, we have been doing this for years, decades. We have figured out where our lane is and how to respond quickly to critical events. Again, I'm not trying to take away from the epileptologist. In fact, these tools help the epileptologist. They're not going to stand at the bedside 24-7 any more than a cardiologist is going to stand at the bedside 24-7. That's us in critical care. We have some algorithms. We interpret the output from monitoring devices that give us complex waveforms that they themselves have been quantified and we take action. Then once we've hit a stable point, we're ready to call in the cardiologist or the epileptologist or whomever else needs to take it to that next level. I sort of hit these three points today that non-experts reading EEG is pretty natural evolution. This is going to happen. I've shown that. That non-experts in EEG are experts in something else, so we can be taught how to see the difference between red and blue and how to see the difference between VTAC and sinus rhythm. By us doing so, it actually helps the experts. Most epileptologists do not want to spend their time in the intensive care unit waiting for a seizure to happen. They really want to dig down deep, figure out exactly what was the cause, where did it happen, what's the best treatment, and then move forward. They really like that fun stuff. That's why they're epileptologists in more critical care medicine. That's where I'm going to stop. My email is at the bottom here and you can reach me there and enjoy the conference. Hello. My name is Aarti Sarwal. I'm an associate professor of neurology at Wake Forest School of Medicine in Winston-Salem, North Carolina. I have the pleasure of talking to you today about my perspective on neurological ultrasound in the future of coma evaluation. I have no relevant disclosures with regard to this talk. For people who are new to the concept of neuroultrasound, we'll be talking about cranial ultrasound, which is an evaluation of patient's brain parenchyma, here seen as an axial section of the brain with opposite skull, midbrain, and circle of Willis visible as color Doppler, as well as transcranial Doppler, where we use pulse Doppler phenomena to look for spectral waveforms and assess the cerebral blood flow and its quantitative parameters. About 10-20% patients have thick skull throughout, which doesn't allow any transcranial insulation using ultrasound, but 80-90% of the patients have a thin part of the skull where when an echo probe is placed, the same probe phased array that we use for echocardiography, also called a transcranial probe, we can insulate patient's brain parenchyma and get an assessment of both the brain structures as well as pulse Doppler for the blood vessel health. Now, when we have much more advanced technology available, why would you use ultrasound as a tool for coma evaluation, which basically centers on evaluation of patient's cerebral perfusion pressure? Well, how about austere environments where other imaging techniques like CT, MRI are inaccessible, or resource-limited environments where they're not available? Even in a subspecialty ICU, there might be a subset of patients, like one seen in the picture here, that may be difficult to transport in case you don't have a mobile CT scanner or inaccessible to the patient's modality at that time. If this patient's coma evaluation centers on assessing pathology in the cranium and their cerebral perfusion pressure assessment can guide direction of further resuscitation, ultrasound can be a great point of care tool. Just like the Mandrukeli hypothesis, coma evaluation also depends on three big pillars. Most causes of coma either emanate from patient's brain parenchyma, which is global cerebral edema or diffuse brain injury, CSF abnormalities as in hydrocephalus, or blood abnormalities as in hypoperfusion ischemia or hyperemia causing hypoxic brain injury, hyperoxic brain injury. We'll use ultrasound to elucidate how these three compartments can be evaluated in clinical applications that can help an evaluative coma. This is an example of cranial ultrasound being used for evaluation of a patient suspected to have an intracranial hemorrhage. In this video, you can see a cranial ultrasound using a point of care handheld device used to reproduce a temporal window. You can see the opposite skull, midbrain, third ventricle, this slit-like structure and hyperdense structure that is very suspicious for an intracranial hemorrhage. Subsequent evaluation by a CT scan revealed a basal ganglia hemorrhage. This is another example of what an intracranial hemorrhage looks like when it's acute. By no means I'm suggesting ultrasound can replace evaluation of intracerebral hemorrhage using a CT scan, but in austere environments or environments where CT is not accessible or available, ultrasound in about 80 to 90% of the patients with temporal windows can help screen for a large supratentorial bleed. Infratentorial bleeds or very small and very high supratentorial bleeds may not be amenable to ultrasound insonation. The sensitivity, specificity and accuracy of ultrasound compared to a CT scan is also not 100%. While it can miss very small bleeds, less than a centimeter or infratentorial, it can also miss bleeds that are subacute or more than seven days old. False positives can be produced by tumors or choroplexis calcifications, which to a novice user can look like an intracranial hemorrhage. Another place where ultrasound can help in evaluation of coma is brain edema evaluation where we're using midline shift. Just like brain trauma foundation guidelines recommend midline shift assessment by using intracranial distance, trans-temporal insonation of the brain can give you ipsilateral and contralateral distances from third ventricle in this case to calculate midline shift. Studies have reported statistical correlations between midline shifts calculated by ultrasound compared to CT scan. Again, in no way it will replace a CT scan, but if serial assessments can be done in a patient that otherwise may not have access to a CT or daily CTs are formidable, this can be a good assessment in a patient evaluated for declining neurological exam or coma. The CSF compartment can be evaluated as well. This is an example of an ultrasound assessment of third ventricle. In the previous examples, you saw a slit-like space that was a third ventricle, but in this case you can see a widely spaced third ventricle that corresponds to the patient's CT scan. And here you can see the choroplexis calcifications outlined by a large little ventricle that acts as a false positive findings like I mentioned in the previous slides. The blood compartment in coma evaluation usually centers somehow on resistance, either focal or global increase in resistance that impairs cerebral blood flow, usually contributes to coma. Cases where resistance can emanate from cerebral blood vessels themselves causing ischemia would be intracranial stenosis, cerebral vasospasm, or vasoconstriction. Places where a global increase in resistance can be contributing to extramural pressure causing ischemia would be increased intracranial pressure. Decreased resistance typically seen in arteriovenous shunting can be seen in peripheral vasodilatation caused by hypercarbia or reperfusion of ischemic brain causing hyperperfusion injury. Spectral waveforms or the shape of the waveforms on pulse Doppler obtained by transcranial Doppler insonation can distinguish these resistance states. On the left are classical high resistance circulation, which typically would be normal, say for a femoral artery, but when present in the brain it signifies very high resistance circulation not compatible with healthy brain. Typical brain circulation when insonated should be low resistance patent, a rapid upstroke during systole, downstroke, gradual and stepwise, leading to a robust diastole, ensuring that the brain gets blood supply throughout the cardiac cycle. This low resistance circulation in the cases of focal or global increase in resistance can change to these waveforms where you continue to see the systolic component, but the increase in resistance has decreased flow in the diastolic component, leading to significant compromise of diastolic flow, which for brain is not compatible with healthy brain. Robust cerebral blood flow throughout the cardiac cycle as seen in this waveform needs to happen for maintaining cerebral perfusion pressure. So if a patient had these waveforms on a TCD insonation, that tells you patient's ICPs are significantly high and are compromising the patient's diastolic flow, causing zero blood flow multiple times during the cardiac cycle and even reverse blood flow. A focal increase in resistance can be seen here. Normal waveforms that transition to high resistance waveforms with impaired diastole and become very resistive where minimal diastolic component almost reversed with small systolic stumps. When focal stenosis is present, just downstream, patient's TCD waveforms can manifest a parvus et tardus waveform. So stepwise insonation of a blood vessel in different segments can show these different waveforms and these waveforms can help you figure out the area of focal stenosis if it's contributing to patient's ischemia in that case. A global increase, say as in malignant ICP crisis, will show a patient with normal waveforms and normal ICP transition to high resistance waveforms where ICP is elevated, but is still allowing diastolic flow. When ICP gets significantly higher than the patient's diastolic blood pressure, patient will continue to have forward systolic flow in the TCD waveforms, but will show reversed diastolic flow. This can be a great screening tool for evaluating ICP in a patient with coma in a non-invasive fashion during point-of-care evaluation. If ICPs continue to be high untreated, they will compromise cerebral blood flow to the point of systolic stumps and eventually cerebral circulatory arrest. The oscillating waveforms classically seen here are considered incompatible with life unless reversed rapidly to these waveforms in real time. On the other end of the spectrum, the waveform at the top is a hyperemic waveform which has normal peak systolic velocities, but very high diastole compared to the systole. Normal diastole in TCD waveforms should be about a third to a fourth of the systolic peak. Here, it's significantly high. This is too much blood flow and is classically called a hyperemic waveform. Where these can be relevant is post-arrest care. Brilliant studies have shown that these hyperemic patterns are associated with evolution to intracranial hypertension and may be associated or harbinger of reperfusion injury after ROSC in cardiac arrest care. On the other spectrum, if a patient post-cardiac arrest resuscitation with ROSC had very high resistance waveforms, nice robust systole, but diastole is significantly compromised. This can also be quantitatively measured as pulse stability index or PI, which is a difference of peak systolic velocity minus end diastolic velocity divided by the mean flow velocity calculated from peak and end diastolic. So peak minus diastolic divided by mean is pulse stability index and this index also called as Gosling's index is a sign of distal resistance and high PI, more than 1.2, denotes resistor waveforms. Brilliant studies have shown that higher PIs or resistor waveforms on spectral Doppler predict an unfavorable outcome after cardiac arrest. This is in compromise to this patient where you have nice robust flow throughout the cardiac cycle but again, hyperdemic waveforms. In coma evaluation with ICP monitoring, we have learned about Lundberg waves and the change in compliance of the brain affecting the ICP waveforms. Those changes of P1, P2, P3 relationship to each other can also be assessed by transient Doppler in a patient who does not have an ICP monitor or cannot have one. The first pattern is the classical pattern we just described where a patient has low resistance waveforms transition to high resistance with increased ICP which may cause reversal of flow in some parts of the cardiac cycle. Some people have used this as a goal directed therapy non-invasively to escalate to surgical evacuation for amenable lesion or osmotic therapy to reverse them to low resistance waveforms. If this patient's brain compliance changes, the P1, P2, P3 waveforms that we see in ICP also reproduced here in TCDs can show a P2 equal to P1 or even elevated and this transition can reflect that the patient's compliance has changed and this patient may suffer a neurological decline at a lower Delta ICP compared to before. People have quantified these changes and serial assessment of resistance using transcranial Dopplers have shown promises in evaluating a patient's cerebral physiology in coma. This is an example from Cambridge of a patient being serially monitored in a global non-compartmentalized model of increased ICP as in TBI where every increase in ICP is manifest as decrease in CPP in a patient who has no auto regulation and these changes can be assessed by transcranial Doppler where the flow velocity on transcranial Doppler acts as a surrogate of patient's cerebral perfusion pressure and goes down each time CPP go down and the PI acts as a surrogate of patient's ICP which is distal resistance and these changes in PI over time as well as cutoffs have been proposed to have prognostic value in TBI and post arrest care. Many people have attempted quantitative assessments of ICPs using non-invasive evaluation by transcranial Doppler but these need further investigation to translate into a point-of-care assessment tool. Another compartment of the brain that has not received enough attention partly because of lag of diagnostic modalities is the venous compartment. Ultrasound can allow insonation of most of the cerebral venous system. This is the transverse venous sinus, basilar venofrazental, often insonated when we're insulating the PCA and your deep middle cerebral veins. The exact patterns venous hypertension or obliteration of these waveforms in patients with increased ICP can be further investigated using ultrasound. So to give you a perspective let's explore a few possibilities of a patient scenario in coma that shows these different waveforms and how TCD evaluation can help elucidate the pathophysiology of coma in that case or the patient stage of coma reversibility and irreversibility. Say a patient had these waveforms, very high resistance waveforms but good systolic flow, positive flow throughout the cardiac cycle but significantly impaired diastole. This is a patient that probably has global increase in ICPs causing high PIs and can benefit if the underlying insult is amenable to therapy from escalating osmotic therapy or exploring surgical options. Compared to this patient where the systolic flow is significantly compromised and the diastolic flow is reversed, this oscillating waveform is technically incompatible with life if sustained and not reversed in real time. A coma patient with normal waveforms, robust diastole, positive flow throughout the cardiac cycle may not have a malignant ICP crisis at that particular time compared to blunted waveforms which are dampened waveforms which reflect cerebral hyperperfusion as opposed to these hyperemic waveforms that could elucidate a hyperperfusion injury in causing patients persistent brain injury and could be a target for therapy. Say in TTM a patient with hyperemia during re-warming may need prolonged TTM compared to a patient that has normal waveforms during re-warming. Venous evaluation could theoretically be a screening tool for assessing uncle herniation and presence of robust basal venipresental waveforms may show evidence against complete uncle herniation. If that's going to guide clinical management, needs further investigation or just an exploratory idea. Patients with VA ECMO have non-pulsatile waveforms but can still be evaluated with transcranial Doppler. So if you asked for an evaluation in a comatose paralyzed and sedated VA ECMO patient, whether patient has significant intracranial pathology, bilateral insulation of the MCS to show robust positive flow could be a sign of intact cerebral perfusion pressure and guide you for the resuscitation. To give you a few case study examples of where I used ultrasound, this is a case study of a craniopharyngeal case that had significant intraoperative blood loss that was washed by thrombolysis as well as saline irrigation but post-operatively had significant neurological decline without focal deficits. CT was unremarkable and so was continuous EEG which failed to show any epileptic form discharges. Transcranial Doppler insulation showed significant vasospasm as manifest in high mean flow velocities 243 and 211 which are much higher than normal of 80 in MCS with resistive waveforms. CT angio confirmed bilateral vasospasm and perfusion deficits and this patient's further therapy was guided with vasospasm with further improvement. Another patient that was a poor grade subarachnoid being monitored for vasospasm had apparently normal flow velocities and normal spectral waveforms with normal PI's, the normal value of that is 0.6 to 1.2. On CL assessment, although the velocities did not change, the patient had significant transition of waveforms to very high resistance waveforms that can be reflected as increasing PI 1.82. Further imaging confirmed, patient had suffered from global cerebral edema that was amenable to therapy. So in this case, neuroultrasound helped in elucidating the pathology of increased ICP in a patient otherwise comatose. Another case study of an intracranial bleed patient that had non-compliant waveforms reflecting high risk of further decline and without change in exam had a transition of spectral waveforms to very high resistance waveforms with reversed flow at some parts of the cardiac cycle. Optic nerve sheath diameter, which is another surrogate for non-invasive ICP, showed high values 0.64 centimeters, which is more than the cutoff of normal 5 to 5.5, and a papilledema surrogate in ultrasound, which is the crescent sign or the optic disc bulge. This together triggered further desiccation of osmotic therapy and surgical evaluation and restored patient's waveforms to positive flow throughout the cardiac cycle, although they continue to be high resistance. Patient had evidence of uncle herniation on CT scan, but intact basilar vein of resenthal. Maybe that showed incomplete uncle herniation, but this case shows how ultrasound guided therapy in a patient that otherwise had no change in exam. On the other spectrum, this is a four-year-old that had H-influenza pneumonia being managed on VV ECMO that suffered from intracranial bleed related to heparin anticoagulation. An ICP monitor was placed and patient was transitioned to high frequency oscillator to guide therapy in a sedated paralyzed patient, but the high frequency oscillations prevented any meaningful ICP monitoring and EVD resulted equally futile results. Transclinodoppler insulation of this patient's MCAs on both sides revealed pretty normal cerebral perfusion pressure, which became a guide for no significant change in cerebral pathology over the next few days. We also used midline shift in this patient as a surrogate for cerebral edema in the known ICH and these evaluations during the subsequent days guided an evidence to continue full supportive care and once high frequency oscillator was weaned, a patient was found to be neurologically the same as before he was sedated and paralyzed. On the opposite spectrum, if the patient had shown these oscillating waveforms, that would have been a sign of significant increased ICP or hematoma progression that could have resulted in either change in high frequency oscillator to get a CT scan and then the mobile CT scan was not available or goals of care discussion. Another case study of an ECMO patient where a sedated paralyzed patient with suspected brain injury was evaluated and showed pretty positive flow in both MCAs, right and left, which gave a good indication that there was no significant focal pathology impairing cerebral perfusion in this patient. To sum it up, I hope that I've made some case to trigger investigations in the role of point-of-care evaluation using neuroultrasound during resuscitation, especially in cases where alternate neuroimaging is inaccessible or not visible and I hope that I've made a few case studies as an example to trigger further investigation in the use of ultrasound in cerebral hemodynamic assessment to add to the understanding of the pathogenesis of coma in different disease states. My own personal quest is that ultrasound becomes a key component in multi-modality guided goal-directed therapy in lots of disease states where we don't understand the pathophysiology or don't have good targeted therapies and it changes our evaluation and management of coma. Thank you for your time. Well, that was a great series of talks. Thank you, everyone, for wonderful presentations. We have had really a lot of interest in this session. And now we are going to do some live question and answer session. And we have a number of questions that are on the question box. Please add your questions if you have not already done so. And we will begin our discussion with some of the questions that have the highest point interest. So Dr. Klassen, the first question is for you. This question is about the recommendation for treatment for lateral periodic discharges or lateralized periodic discharges that are greater than 2 and 1 1 that you mentioned in your lecture, even if these do not have an inictal evolution. How should we treat these? Should these be treated aggressively as if they are seizures? That's what the question is. So Dr. Klassen. Great. Thanks, Paul. So that is a great question. It sort of hits at the core of this ictal-intraictal continuum. And what do we do with this? And I think initially when we started this field of EEG monitoring, we thought it's going to be more straightforward. It's going to be there's going to be seizures, and there's going to be normal brain activity. But clearly what has emerged is that there's a large, large area of patterns that are somewhere in between. They don't classically meet criteria. So the American Clinical Neurophysiology Society has spent a lot of efforts and should be really recommended for that. They spend a lot of effort trying to define first what we're looking at. So developing a catalog, a sort of in some ways language, how we communicate about these patterns. And so one of the patterns that has emerged is lateral periodic discharges, PDs. And what the data shows is that if they are at a very high frequency occurring, so above 2.5 Hertz, they behave pathophysiologically similar to seizures. And so a lot of people, including myself, tend to manage them similar or the same as electrographic seizures. So that's, I think, what we now do. And so I think the ACNS is basically following the field and is now also recommending that. So I can only agree with that. Great. How about our other speakers, Dr. Sarwal, Dr. Olson? Any perspectives on this kind of treatment? No, I think I just echo Jan. I think we have followed the current recommendations and we need a little bit more understanding at the bedside of these implications. Actually, Dawei makes an important point. It's not a static field, right? So I think this is where we are at this point in time. And I think we all should be open for, first of all, pushing for getting more data, understanding these patterns better, trying to see whether treatment actually really helps doing clinical trials to support that. But I think at this point in time, that's basically we would lean towards treating them the same. Great. Great. Let's move to our second question. And this is a question about training and empowering intensivists in the ability to use some of these quantitative methods. And how do we engage our colleagues? What are maybe some best practices or some commonly used practices of making the EEG for everyone, the non-expert, especially the quantitative EEG segment? So Dr. Olson, perhaps you can, since this was the title of your talk, perhaps address this first and hear what the others have to say. Yeah, I'm excited to hear what everybody else has to say. I think the number one thing we need to do is exposure, exposure, exposure. It's always fascinated me that people are like, I don't know if I'm going to use this. I don't know if I'm going to use this. It's like, oh, the Apple Watch or any device or any technology. And there's sort of this weird resistance against trying a new thing, right? I don't know if I'm going to like broccoli. And then they try broccoli, and they go, shit, I've never tried broccoli. Broccoli is amazing, right? And I think as intensivists and nurses, as we get exposed to seeing this technology, as we start seeing more and more EEGs, I think that's what's going to turn the tide, is you need to start using it. And you can't wait until you're absolutely 100% certain that this is the patient in whom it will make a difference. Because by then, it's too late. That's not the patient you learn stuff on, right? We're not going to wait until someone has the cardiac arrest to practice CPR. We're going to build a dummy, and we're going to practice our compression. So we're ready for the real event. So I think my answer to this question by Dr. Diaz Gomez is, we've got to start looking at it. I don't know. Dr. Sarwal, what are your thoughts on this? Yeah, I think, the way I hit the point, home, my biggest motivation and efforts by a bedside is usually to engage people, especially when I'm doing these consults in the medical surgical ICU where a knowledge and appreciation of these tools is not enough. We really need more data on our population. We really need engagement of the clinicians by bedside to integrate that data into patients' clinical picture and understand the pathophysiology of the patient. I'll give you a little example. I teach bedside EEG to any nurse who's willing to learn. When you see VTAC, VFIB, when you see VTAC, that's a bad sign in EEG. And when you see coarse VFIB or just squiggly, wiggly line that's a good sign relatively. So I think, and the nurses are willing to learn. And the wife did a really kind of energizing and humbling statement. We all are experts. And I think engaging our non-neuro colleagues to the table to help them learn when and where these tools can be useful to make them your champions is going to progress our field and help our understanding of better care paradigms for our patients. Yeah, can I add something to it, Paul? Please, go ahead. So I couldn't agree more with both Artie and with Dawei. Very well said. I think there's a couple of things we have to realize. One is seizures, period discharges, abnormal EEG patterns are not limited to neurologically injured patients. And we've shown, we've and you've shown and other people have shown that these seizures can be seen in patients in a medical ICU, in a surgical ICU, in any critical care setting. And so I think with a paper by Town, I mean, in the last decade and last century, actually in the late 90s, he already showed that electrographic seizures are common in the medical ICU. So I think we have to, by that understanding already, we have to reach out. And I don't think we should underestimate the ability of people to learn technology and techniques. The other thing is that a lot of even brain injured patients are not managed by neurocritical care specialists. Many, the vast majority of them are managed by anesthesiologists, by pulmonologists. So I think it is up on us to really reach out and empower people with the knowledge to really understand, get a basic understanding of the EEG. So I think I couldn't agree more. And I think what's important is to not make it sort of protected and make it seem more complicated than it is. There are some basic things that you can understand. And there may be ways where you can call in a consult when it becomes a bit more difficult and you don't understand it immediately. So I think both raw EEG and quantitative EEG can be actually brought out to any critical care setting. Yeah, and I would add that there is some literature on this. Dr. Swisher published a very nice article on exactly this point of training bedside clinicians, nurses, and physicians both on the use of quantitative EEG and then comparing that against kind of the electrophysiologist recordings. And there was very good, there were very good correlations, very good agreement between these various types of users. And that may be a type of article to sort of read. But the process that they describe there is really one of training, addressing it, addressing how it can be used, and engagement. And so training courses are one ability to do that. Training in real time, case discussion, we always learn during case discussions. All of our colleagues, no matter what the subspecialty is, both do a consult, but also provide education. And so that's a very common way to another potential possible approach. I, Daiwei, I think you were going to- Yeah, yeah, I just want to circle back to get your, everybody's opinion. So, you know, nurses are not as disease focused as we are systems and process focused. So, Yan, when you said that even in patients who don't have a brain injury, but in my mind, if I am so sick with my medical diagnosis that I start seizing, isn't, I mean, at that point, don't I have an injured brain or, I mean, how are you teasing this out as to, is it just location? No, totally, Daiwei, you're totally right. And I mean, you know, honestly, I mean, a lot of the patients, there's a higher rate of electrographic seizures in septic patients, which are often in medical ICUs. There's a ton of data now showing that actually patients with sepsis do have brain injury, right? And you are totally right. That's a matter of, I was talking more in terms of like, let's say turf of where people think this is a brain injured patient and this is a non-brain, but you are totally right pathophysiologically. I mean, most of these diseases are whole body diseases, multi-organ dysfunction diseases, right? So of course the brain is not separate there. You're totally, completely right, 100%. There's another question that's related to this. And then I think we should talk a little bit about some of the other topics, but the other question that's related to this is one of building sort of a neuromonitoring program and retention of technologists and retention of, you know, the challenges of doing this from a practical logistical perspective. And maybe, you know, what are some lessons learned across our experts here of how you've made it work in your center? So it's tough. It's a huge challenge, right? Especially if you live in an expensive city like New York City to retain people is actually not easy. And technologists are one of the, you know, sort of bottlenecks to put a system like this together. So absolutely right on the money. The one thing we did at some point, we actually founded a school, a technology school. So that actually allowed us to have quite a lot of, you know, people that came in. And then the people that were with us actually became educators as well. And I think it is, I think what I always think about is it's about empowering people, right? So one is either as educators, for example, as a technology educator, or I have actually always at least one or two of our EG technologists involved in my research. And so some are more leaning towards an education path. Some are more interested in more like a research path. And I do think that we have to think beyond just physicians there. And this is something where in my, basically, in my research team, for example, they have physicians assistants, several in their nurses. It's actually a multidisciplinary team in that sense. And I think that helps keeping people engaged. And then the other thing is also to cherish the data and discuss the data, make people basically in rounds to discuss what you're seeing there, go over that specific patient that they're taking care of. And so that makes it relevant then, I think. Great, Dr. Olson, Dr. Sarwal, are there perspectives on this? I think for us, this has been the big challenge because we don't have access to an EG technology school, but that's a great idea for me to start working towards. We do have some technical community colleges. What we have done to counteract that is start using point of care EG devices. Devine mentioned a few ones where we actually supplement our point of care EG, or defined EG that's placed by nurses. And we have algorithms of when we escalate to continuous EG for full 12 hours versus 72 hours. And again, Han did a great job of saying, you have to engage them, you have to empower them. So our EG techs are very engaged and empowered to actually work with the clinicians to work with the clinicians. To actually work with the clinicians to understand the urgency of EG machines as well. So rather than just doing every EG stat, they actually communicate with us and we have a very nice multidisciplinary clinical kind of text paging system where they try to triage stuff. So we work with them, but again, I would love to have a 24 seven robust program, but the challenge is real. Great. So the next question revolves around the waveforms of ICP monitoring. And I think in turn is similar to the transcranial Doppler waveform. Traditionally, the ICP waveform has P1, P2, P3, and you can learn something about compliance. The question generally is, are we using these compliance methods, any provocative tests perhaps to evaluate for compliance, both either from ICP monitoring or from the transcranial Doppler? So Dr. Sarwal. So great question. I think I'll start by the caveat, we definitely need more investigative research. And that's one of my biggest motivations in sharing at least some of the data and one of my biggest motivations in sharing at least my experience in using such a handy non-invasive tool. We have lots of neuro ultrasound opportunities for people to learn. TCD waveforms are essentially extrapolation of the pressure changes that occur in the brain and the transitions of P1, P2, P3 can be reproduced just like they come on ICP monitor. I personally have not done provocative tests partly because the first thing is always do no harm. I think I do enough point-of-care TCD serially in a patient where I'm interested in understanding changes in compliance that I will perform serial evaluations myself, even if they're just brief MCA insulations, like the ones I showed you in the ICH example, the non-dominant ICH where, you know, patient's compliance changed significantly, although the exam and the CT scan did not. And that gave me a very heightened sense that this patient is more likely to deteriorate and it would require a lower delta of ICP increases to deteriorate that patient. So this patient actually deteriorated from an ICP of nine to 15 at the time when ICP waveforms are non-compliant as well as the TCD waveforms. So they can be done, but I think I would suggest people to do more investigative research and reproduce this in clinical trials to make sure we're not doing patient harm by doing provocative tests. But non-invasively, I would highly recommend everybody to use TCDs whenever they don't have an ICP monitor, exactly the same way they use the ICP monitor for P1, P2, P3 correlations. And Dr. Sarwal, a couple of questions have come up related to that, and they center around the time window or temporal changes that may occur after a certain clinical situation. So the first question is pretty generic and is how often do you do neuroultrasound? And perhaps you can start with that one, and then I'll ask the second question that's more specific to a disease state. So my real honest answer is as often as time and opportunity allows. Sometimes because I want to know more about my patient's physiology. Sometimes I just want to look because you never know what you find until you look. And as my example from the Ross patients showed, sometimes I find hyperemic waveforms, and sometimes I find high resistance waveforms, and that completely changes how I'm going to target this patient's osmotic therapy versus blood pressure management. Physiologically, I think patient is auto-regulating, but the TCDs argue otherwise. And that makes me sit back a little bit and revisit my clinical physiology-based hypothesis on that patient. So I take every opportunity a patient and my schedule allows to look into the brain, especially having such a non-invasive tool that is not harmful. Yeah, you mentioned a really good point, and that is evaluating the TCD, having a hypothesis, and maybe changing your therapy. And there have been publications on auto-regulation testing and doing a baseline TCD, then making a step change in blood pressure or some other parameter, and then repeating the transcranial Doppler. How often are you using it in that kind of fashion, looking for a response to your therapy? So I do this in any patients that have a suspicion of high ICP as often as I can. Like I told you, I actually use these waveforms as a goal-directed therapy. So if the patient is a candidate only for medical therapy, I will escalate medical therapy until I get their waveforms to a positive net forward flow. In the ICH case, we actually use this to trigger surgical evacuation. And post-evacuation, we're able to see positive waveforms. And on the corollary, in post-TTM patients, where they're going into a reforming phase, I actually use the hyperemia to make their blood pressures normotensive. Because our colleagues tend to say, let the blood pressure ride where it is. But my concern is, I need more data to prove that. And hopefully, the pre-ISCAP trial will help us with some TCD data. But these hyperemic waveforms may represent reperfusion injury in these patients and may be causing the parenchymal injury, even if they don't progress to cerebral edema. So I think there's a significant value in using these waveforms. If you have the expertise to follow qualitative waveforms, great. The reason I propagate the whole spectrum is basically to empower our ICU colleagues. If your skill sets allow you to only have a point of care ultrasound that only gives you basic Dopplers, here's a way to look at them. And here are the patents. If you have advanced AI software or machine learning software, or even processing software like ICM Plus, and you have the capability of having parameters like PRX or MX to follow the same thing, this is how they look. So depending on the spectrum of your expertise, technology, and infrastructure available, TCDs can be extremely useful goal in the goal-directed therapy. That's great. And a related question comes in terms of the specific application of transcranial Doppler to cardiac arrest after ROSC. And the question is, what are the time windows for the various changes in resistance in cardiac arrest patients after return of spontaneous circulation? So that's a great question. I really don't know the answer. The answer is I don't know, because serial studies have not been done. But knowing how enthusiastic I am to look at these patients' brains every time I'm consulted, so all our TTM patients have a mandatory neuro-IC consult. Anytime I'm on, I have TCD data on these patients. What I have seen is almost a trichotomy of patients. So there are three sets where the patients fall. Some patients have such significant early ischemic deficit that they already have high ICP global cerebral edema. And the next 40 to 72 hours are going to progress no matter what you do in these patients. These patients typically have very high resistance waveforms. The validity of high osmotic therapy and other ICP learning therapies is still to be investigated. But at least in the current paradigm, these patients don't survive. The second set I see is patients that have reperfusion or hyperemic patterns early on. And in these patients, there needs to be investigations whether blood pressure lowering would actually help reperfusion. These are the patients probably auto-regulated. These patients fall into further two subgroups. One set of patients get reperfusion injury and actually develop cerebral edema in a delayed fashion, typically around day four or day five as I follow them. And then the second group never really develops delayed cerebral edema. But they get significant brain injury that is parenchymal and manifests on MRI. So in my N of 20 that I've collected the data on, there seem to be these three groups that are coming around. How to use that data to actually change outcomes is what needs further investigation. That's fascinating. You know, it's fascinating to consider sort of the evolution and change in cerebral hemodynamics from the TTM period to the post-TTM period. And it's just fascinating work. That's wonderful. That's wonderful. A related question that comes is, have you used transcranial Doppler in patients who are on cardiac assist devices like LVADs? And there's some related questions to that. Are there any crucial research findings that are coming forward on that? The big overwhelming answer is yes. My cardiothoracic anesthesia colleagues are actually very progressive in endorsing non-invasive strategies. We don't have a mobile CT scanner. So I am very closely involved with evaluating patients with BA ECMO, VV ECMO, Impella devices. And I actually give a full hour lecture of some descriptive case series on that. The few things where I find these useful is number one, screening patients for high ICPs. Since we don't have an exam on these patients, many of these patients, if properly selected, may have a cardiac arrest that may impair the brainstem for 24 to 48 hours. And considering what a big resource utilization ECMO utilizes, early on, identifying a patient that has very high ICP patterns can be a very useful information for the rest of the team to understand the multisystem organ injury and incorporate that into the prognostication algorithms. So one is screening for high ICP and malignant crisis. The second is screening for focal pathology. Even in BA ECMO patient, where you see non-pulsatile waveforms, you can see distinct asymmetries of more than 30% to 50% that can reflect focal pathology if you combine that with cranial ultrasound. So patient is anticoagulated, has blown pupils, was eight hours into cardiac arrest. Your TCD waveforms show beautiful, normal, mean flow velocities bilaterally. At least patient does not have a focal intracranial pathology impairing cerebral perfusion. Versus another patient that has a non-pulsatile waveform on the right side with a velocity of 80, but on the left side it is 30. And you see the remnants of what may be an ICH on the left side that gives you a pause to at least take the high risk intervention of getting a CT scan on this patient. So these are the two places where my colleagues find my point of care assessment very useful. The third place where I would like to see people do further investigations is most of these patients have some degree of injury, whether it's anoxic, ischemic, or whatever hypoxic, what led to the initiation for the assist device, whether it's ECMO or Impella or other bypass procedures, or even intra-op for that matter. What I would like to know is, I suspect that these brains are dysautoregulated. And I wonder whether this dysautoregulation in a chronically hypoxic brain or acutely injured brain is predisposing them to hyperperfusion injury and that hyperperfusion injury is affecting cognitive outcomes. If you look at the post-op literature on these patients, up to 50% of these patients either suffer from a neurological event, which could be seizures, which is more often ICH, but is very often just delirium as well. And these cognitive deficits last for years and months after their cardiac event has happened. So I wonder if reperfusion injury is contributing to injury in a dysautoregulated brain that has heart failure or underlying cardiac insult. So that's where I would like to see a research agenda expand and more involvement of a cardiothoracic people involved in the care of these patients. Wonderful, that's great. Dr. Olson, Dr. Clausen, any other perspectives on transcranial Doppler ultrasound in use in your practice, both in the cardiac type patients or even in the neuro patients? Sadly, I think Jan had to run for a COVID emergency and I'm not sure if he's gonna make it back on. Dr. Hamlin asked a fascinating question to me that I think relates to the whole use of the point of care. And I do remember when we used to compress IJs or push over the liver to watch the P1, P2 change. One of the things in terms of a design for a research study, and it would be observational, is we have a natural experiment of a lot of these intracranial dynamic changes, both with that can be seen on EEG and on point of peril. So there's a lot of nursing interventions, things that we do that are known to associate with ICP. When we lower a patient's head of bed so that we can roll them from side to side and clean them. When we suction, there's lots of natural events that are unpredictable in terms of when they're required or when they will happen. But this also speaks to an earlier question someone had asked about retention. So imagine that the overall experiment is we're teaching our nurses, we're teaching our APPs how to get point of care ultrasound transperineal. So it's much more readily available. First off, I can tell you as a nurse, the more you teach me, the more I'm staying, right? I wanna be continually growing in my professional practice. So this is yet another instrument that I can use or tool that I can use to understand what's happening with my patient. But this may also help folks like Dr. Hamlin get this data that they want, right? So you can imagine POCUS before and after suctioning. And if you've got EEG on, and if you happen to have an ICP, I mean, you can really pull this information together. And that's how I think we're gonna move forward on some of this. So the statement's been made again and again, we need more data, we need more data. Well, the data actually is there. We're just not collecting it. Yeah, and I think it's one theme from both EEG and intracranial pressure and transcranial dropper ultrasound, neuro ultrasound is that the brain is very dynamic and there are moment to moment changes and that can be seen at the bedside. And that as information can be displayed at the bedside and seen, you know, observed by the nurse and other care providers that are at the bedside, observations can be made and feedback be given. Even if those observations are not about seizures per se or, you know, a critical event like a vasospasm or, but they are these kinds of dynamic physiologic effects and integration of information occurring at bedside. So one of the themes is that really bringing the information to the bedside is important and integrating it there is an important part of this field, I believe. There's a question about EEG software that has come forward and the question is posed, are there new software packages to simplify interpretation of EEG similar to radiology where the computer is sort of doing the work? And Dr. Olson, you began mentioning some of this and perhaps you could elaborate on that. Yeah, without, the answer is there's an increasing number of software packages coming out. I don't think we have coalesced yet on what the package is, nor I mentioned GUI. We don't yet know where we're gonna land on this. I think it's kind of an exciting time to be involved. The hard part is, do you buy a package that you stuck with for five, 10 years? What's the training, the learning, you know, the learning curve? I'm not gonna get into any specific company for obvious reasons, but there's a lot of packages out there what I, but in preparing for this talk, I tried to find out like, what are the pros and cons? Have they been compared? Not as to is this one better or worse, but does this one meet my needs more than this one meets my needs? Because I happen to be a general intensivist at a non-teaching institution where I'm running a general ICU that gets everything from a pediatric to a geriatric and every diagnosis, that person needs a different package than the neural intensivist at a university setting. So short answer is, yes, there's a lot of packages and there's gonna be more next year. And there's a related question about machine learning and artificial intelligence. And the question is, what do you think the role of machine learning artificial intelligence is for continuous interpretation of continuous EEG? Perhaps you could talk about that. So I think other people have heard me rant on machine learning. My humble opinion is machine learning is multivariate modeling. It's a sexy name for multivariate modeling. We've had this before. The computing, the computer power we have now allows us to do something a little bit different but it's really taking multivariate modeling for it. So if you're asking the question of, does entering more variables into the model help us better predict? The answer is yes. So I think machine learning, however you wanna call it, machine learning is going to be part of EEG interpretation in the future. But in a very real sense, it's multivariate modeling has been around for a long time. I know Dr., I think Dr. Bleck was commenting earlier. So I suspect he's on. And he's sort of my go-to for history of how we've developed some of the statistics. I'd love to, if he typed into the chat, hear his comments on this as well. What do you think, Artie and Paul? I think the amount of monitoring data that has been thrown at us and the expertise and the gist of people that are available to process this needs some extra help. Now, whether the help comes in the form of machine learning, AI algorithms or processing softwares, how you call it, I think there definitely needs to be a collaboration with biomedical engineers and data science experts in helping us interpret this information more often. Yeah, I think you're right. I think that there are a number of software adaptations for continuous EEG that have been coming forward. The quantitative EEG displays that Dr. Clausen was showing and Dr. Olson was showing are ways to sort of graphically show long periods of EEG that call for a deep dive into it. Why has that change occurred? And the same type of thing, actually, Dr. Sarwal was talking about, taking the transcranial Doppler or the intracranial pressure monitoring and putting it through software that trends it over time and then looks for relationships between parameters, correlation coefficients between certain parameters like blood pressure and ICP or blood pressure and the transcranial Doppler velocity. And those have been the very first, those are sort of graphical displays, if you will, or user interfaces that add value, add a perspective of what's happening over time. And those are the things that are sort of commercially available. They are still labor intensive. They require interpretation and training. And then over time, we probably will see more informational algorithms and that type of technology come forward. That all of these quantitative methods are used oftentimes in dual display with the raw EEG. And I think looking at the raw EEG is going to remain important as well. Dr. Black has a question about nursing protocols and whether we're at the point where nurses are using protocols to initiate and alter treatment based on the analysis of EEG data. Perhaps Dr. Olson, you could answer that and Dr. Saru too. So this is obviously near and dear to my heart. We are not at the point where we are doing this, but we're at the point where we're ready to do this. When I first started my practice well over a year ago, back in the early 80s, not all nurses were allowed to read EKG. You had to go to a three-day class, right? And then somebody came up with this crazy idea called ACLS. And I know Romer was heavily involved in this and some other folks. And the hospital that I started at, at night had one physician in the emergency department, no other physicians in the entire building. So someone had to respond to a cardiac arrest, run up to the floor and nurses were there. So as ICU nurses, we ended up running codes and making decisions. This is VTAC. Now, granted back in 1983, we gave Bi-CARB. We've learned since, but we had algorithms, right? And we had figured these things out. I think we're ready for that. We're ready to take ENLS to the next step and respond to seizure. We're able to respond to increased intracranial pressure. Our advanced practice providers are licensed to do that. As an RN with ACLS, I'm licensed to give epi in certain situations. Where I picture we're at is, there's the great nature documentary that shows a whole bunch of penguins marching down and they're at the edge of the water, just kind of, everybody knows there's a whale down there somewhere and no one wants to be the first one to jump, right? I think some brilliant university is gonna be the first one to jump in and allow nurses to practice to the fullest extent of our licensure. And then everybody else is gonna go, well, they did it at Yale, let's go. And I think we're ready. We've got the knowledge. So Dr. Sarwal, Dr. Vespa, are you with me as UCLA? Yeah, Dr. Sarwal, what's your perspective on this? Well, while we're waiting for that, Dr. Bleck did respond saying that we are working on ways to use machine learning for ICVG analysis, but one stumbling block is the lack of a gold standard for non-convulsive seizures and status. It's hard to teach a model when you can't verify the outcome. Hopefully this will solve the problem, but today I don't know how. So yeah, I agree with that. That's very interesting. A different question comes about a transcranial Doppler ultrasound. And this question is about, is there any point-of-care ultrasound for transcranial Doppler data in patients with diabetic ketoacidosis? Dr. Sarwal. That's a good question. I think I would love to gather that data. I'm not aware of it. I can probably Google it real quick. This is not a population I've typically been involved in when we incidentally had diabetic ketoacidosis in your ICU. Thankfully, I haven't had a patient that had coma or kind of ICV crisis, but I would encourage you to learn some neural ultrasound. It's not difficult and explore that. I mean, who knows what you'll find. Great, great. There's a question about developing a skillset for 24-7 teams and to be both in ultrasound and in continuous EEG and other kinds of monitoring. The question is, should a combined competency pathway be created for physicians and nurse practitioners or other practitioners for developing skillsets in this type of monitoring? I think I would say yes. And point-of-care ultrasound has done a phenomenal job in paving a way for that. It doesn't matter whether you're a nurse, an NP, a PA, a surgery resident, or a neuro-intensive care physician. If your patient is in shock and you know your four chambers and three causes of shock, it can be easily taught for you to get the four basic views and help understand the patient's physiology of shock based on those views. Obviously, there needs to be some training competency, but you teach it to a nurse versus teach it to me. Getting those four basic views and correlating that to your understanding of shock and anatomy of the heart is pretty much the same. So I absolutely wholeheartedly support that and willing to work towards that. I'm kind of with it, but I think it would be a very limited combined competence. I think there's certain skillsets that we want to share. And again, I use the analogy of BLS-ACLS. We share the skillset of defibrillation, but there's, quite frankly, as a nurse, I have a different set of training and a different perspective that I find is unique and valuable. There's a different skillset that physicians have. And I'm perfectly okay with that. In fact, vive la différence. That's how we work together. So I think the combined skillset, I'm sorry, I couldn't find the question, so I'm trying to remember it. The combined skillset would be limited just like it is in ACLS, right? My skill in ACLS doesn't extend all the way through doing a PCI. Right. But I know what's going on, right? I know we're going to move to this. So I would be pretty hip for some competency that really models after what we've done with ACLS. I don't know if I really answered the question. And Deva, you had a beautiful slide that portrayed, you know, we're not expecting a person to sit at the computer and say, you know, this patient had a three-second focal seizure emanating from the right temporal lobe and generalizing. You literally, those competencies will focus on saying, hey doc, this patient's not doing well. I just did this test and some, we really need to look at that. So I think having that combined clinical competency of certain basic ABCDs, using neuromonitoring tools that actually help raise a concern for escalation is where I see this combined competencies be aiming at, because right now we don't know what we're missing. We don't know how many triggers are we missing. You know, not everybody has the comfort and the luxury of a 24-7 invasive neuromonitoring program like some centers here do. But that doesn't mean that the 90% centers out there can't gather that data and add to the science in different ways. Right, and I think that as we were talking about before, that displaying information from multimodality monitoring at the bedside in real time, which these devices do, enables nurses to be constantly monitoring, identifying abnormalities, identifying worrisome abnormalities, engaging their physician teams for interpretation of data and discussion about what the nurse's interpretation of the data is. And that is a way, I think, to really bring forward this concept of continuous monitoring of the patient. It's a team approach and competencies based on the particular goal of that team member. And nurses have, Daiwa, you can speak to this, but there's a long history of nursing competencies and all of the different kinds of monitoring techniques that we do and many other things. Perhaps you can speak to that, yeah. Well, yeah, I do want to speak to that, but you brought up a really interesting point, Dr. Vespa, and this circles back to something you said much earlier. As a nurse, I am really intensely monitoring the patient for 12 hours. I am minute to minute. My epoch, my window, my temporality is 12 hours. And you had mentioned earlier trending. So how do you think, how does trending, and what's our temporality window that we should be looking at or how, and how does that play into competence? Yeah, I think trending is important and many of the displays that we have are trending. Many of, you know, the EEG for sure is doing that. Some of the ICP data, brain tissue oxygen monitoring, there are trends that you can see. And obviously the charting. Nurses are, and nurses are charting or the systems are charting it semi-automatically and the nurses are constantly evaluating trends. I mean, this is, and physicians are too. And I think in intensive care, one of the principles is that we are looking at temporal trends of data, that the individual, you know, cross-sectional time point may not be as important as the trend of what's happening. If you're watching somebody's intracranial pressure hour by hour and it's going up, well, that's an important trend. And I think that we have developed competencies to evaluate both the individual number and the trend. I think that the challenge perhaps is, as Dr. Sarwal has outlined, integrating the dynamic waveform change with the trend, with sort of the recent historical baseline for that particular patient, and then interpreting what that means. And I think that that's where it really becomes a team effort of really trying to understand, for example, that TCD is changing because the temperature is going up and we're rewarming this patient in the context of now their blood pressure is, their mean arterial blood pressure is, you know, 20 points different than it was before. And integrating that information in a truly multimodality sense is a group and team effort, I believe. Dr. Sarwal? I think I couldn't have said it better. I think you really hit the nail really well here. Yeah, I couldn't have said it better. Let me go back to the questions. I think that we have addressed most of the questions. There are some papers on training. Dr. Ellender has put forward a link to a paper. I think this was perhaps mentioned in one of the earlier talks in terms of bedside training to recognize patterns, important patterns. There are several papers like that in the literature that Dr. Swisher was the senior author of one of these papers that specifically addressed quantitative EEG and the use by bedside clinicians, non-expert bedside clinicians, if you will. Other comments? Let me see if there's any going through the list here. Lots of compliments to our speakers from the groups, from the group that has come through. Wonderful accolades, just wanna pass those on. We're nearing the end of our timeframe here. I think that, so if there are any last minute questions or comments, please feel free to type those in. There was a comment about a study on TCD and, excuse me, cardiac assist devices. I don't know if, I can't, there's an article that was quoted here about emboli monitoring. But I don't know, Dr. Sarwal, if you, wanna comment on that. I haven't had time to look at that. Definitely, so I think that's another area in coronary bypass that has been explored very nicely. People have used emboli monitoring to optimize surgical approaches in cardiothoracic bypass. Many surgeons have integrated neuromonitoring since cerebral emboli are a huge real-time marker here and do correlate with post-operative strokes in these patients. The other place where in VA and VV ECMO, I don't, again, by any measure, recommend that TCD should be used just for this purpose. But we have incidentally, while incidenting a patient, found embolic hits, and that has led to the perfusionist understanding that either there is air in the circuit or the patient has clots and so on. And the heparin needs to be escalated. So there are very many ways, I think, neuroultrasound can be used in assist devices in addition to the research emphasis on understanding the role of reperfusion that I mentioned earlier in the discussion. Great, great. There are some housekeeping comments here. The questions that were raised here will be viewable through the remainder of the event. And there's also a comment about CME information. That's on the homepage for the SCCM event. And I think that's all the questions that we have. We're gonna remain available here for the next few minutes. If anyone has anything else that they want to bring forward. I just was, that there's a little bit of a delay in the stream, so we'll give it a little bit of a timeframe here and check back in just in case somebody has a last minute question. Dr. Vestma, I'll ask you one last question. You host the ICU of the future meeting, so what's the future regarding these technologies? Yeah, well I think that the, my sense is that the future of multimodality monitoring is going to be devices that probably have some more technology built into them, some of this automated technology. But I also think that really the future is going to be the use of this really probably in more ICUs, I mean what we've seen published in the literature is the fact that the non-neurological patient who's critically ill has a lot of neurological complications. The sepsis patient has a sepsis-associated encephalopathy, perhaps a brain injury of some kind. Those patients are being increasingly monitored, the work by the Columbia group many years ago and others now show that. The work in cardiac arrest, the work in other populations of critically ill patients show pretty high incidence of seizures, pretty high incidence of epileptiform discharges that potentially are pathological. So I think the ICU of the future is going to be probably the propagation of this concept really beyond the traditional neuro-ICU setting to general medical and surgical ICUs. So I think that's probably one of the biggest methods going forward. I do think that technology probably will continue to precede these kind of machine learning or other kinds of technological solutions for both graphical display but also interpretive display, interpretation of the signal. So I think that those are going to be pretty important. Okay. This was a great session. Thanks, Dr. Sarwal. Thanks, Dr. Vespa. Yes, thank you all. Thanks, Dr. Claussen, who had to sign off. Thank you, Dr. Sarwal and Dr. Olson for a wonderful session. I think I don't see any additional questions here. So I think that... If I can make a shameless plug as a last thing, please consider attending the session next Friday, Stump the Intensivist, where three eager intensivists talk about three different modalities, and I will be arguing the pros and cons of using neuro-ultrasound in the evaluation of coma, and we'll see if we can stump you either way.

brain monitoring techniques

ICU

EEG

seizure detection

prolonged monitoring

automated EEG

quantitative EEG

non-experts

comatose patients

collaboration

neurological ultrasound

cranial ultrasound

transcranial Doppler

brain activity assessment

resource-limited environments