Hello, my name is Alicia Everett and I will be speaking to you today on detecting secondary injury in real-time using bioimpedance in a pig model. I would like to thank my co-authors. The outline of the talk today will go over TBI, our technology which is electrical impedance based, the methods we used in this study, our ability to detect intracranial mass effect in our future directions. We have nothing to disclose, but this project was funded by these grant sources. So obviously TBI is very prevalent and a huge problem in our society, specifically it's the fourth leading cause of death in the U.S. and even among hospitalized patients mortality can be as high as 20%. If you suffer severe TBI, one of the gold standards for monitoring of these critical patients is to implant an ICP monitor. However an ICP monitor helps to monitor the status of the patient, but is limited in its ability to detect secondary injury. Specifically to discern if the secondary injury is due to something focal such as a contusion or a cerebral hemorrhage or something diffuse such as edema or hypoxia that could just be treated with mannitol at the bedside. In fact elevated ICP could be due to a variety of underlying etiologies and in fact identifying the etiology behind hypertension is imperative to proper treatment and patient recovery as well as prediction of that patient's response to any future care directions. What is electrical impedance? Impedance is the complex resistance of a tissue to current flow. So within this you can apply a current at several frequencies and that frequency is going to take a different path through tissue based on the state of the tissue as well as the path due to the frequency that you apply. Impedance can be represented by these equations shown here. I will refer to it as Z in this talk and it's the sum between the real and imaginary components of that tissue's resistance. It can also be modeled as several varying complexities of parallel RC circuits which helps in differentiation and identifying different tissue pathologies. You may already be familiar with impedance. It's used in BIA, lung monitoring, and it is the metric utilized in many familiar devices including cautery and vessel sealing. So why did we choose to tackle this problem of secondary detection and differentiation with impedance? So specifically it's very sensitive to the pathology state or pathologic state of tissue. In fact it's already been shown to be sensitive to blood changes as well as tissue ischemia. And while ICP responds to pressure, impedance can be sensitive to volume. And this is really exciting because volume is going to be able to be representative of the conductivity that's present versus just the pressure state. You can imagine that we would have a setup much as shown here where there's peripheral electrodes circumferentially around the head and one central electrode and we can change where we're investigating in the brain by which electrodes we drive current across. This is going to allow us to, while pressure might increase for both a ischemic and hemorrhagic event, this allows us to use impedance to potentially differentiate between these two which is our hypothesis because while pressure increases impedance responds oppositely. How we investigated this was in a miniature pig animal model. We investigated two established models, one with a high impedance impact which is our ischemic model that consisted of inflating a Fogarty balloon and the other was a low impedance model for intracerebral hemorrhage which consisted of injecting autologous blood. The instrumentation looks like this with three intracranial catheters that were placed, one with a compliance balloon into the left lateral ventricle, a coupled ICP and intracranial electrode went into the right frontal lobe, and then our mass effect for that focal intracranial volume went into the left parietal lobe. This is what it looks like once there are surface electrodes and full scalp instrumentation. We used surgical guidance to be able to be precise and repeatable between all animals. Axiom stealth, registration, fiducials, and planning permitted this level of control. For the surgical protocol we hand drilled burr holes and placed plastic tui borests and intracranial bolts to avoid any artifacts in the CT. The protocol consisted 30 minutes resting at baseline followed by a balloon inflation in steps of 100 microliters each every five minutes up to a total volume of 1.2 mils. The balloon was then deflated in the same procedure and autologous blood injected in steps of 200 microliters every five minutes up to the same total volume of 1.2 mils. We then treated with mannitol and after 30 minutes euthanized. Simultaneously at every volume step we were collecting CT scans. What we see here is preliminary model validation. These are CT segmentations where you can see the placement of the burr holes and further in yellow that is an inflated Fogarty balloon. In C of the image on the left, those are validations of the Fogarty at two different volumes, one zero milliliters and one in 1.2. By mixing contrast into our saline that we were injecting into this balloon we were able to image it in real time as well as post-mortem to verify and validate our volume control. On the right side you can see also a visual of the blood being injected right at the tip of that catheter. It went through the thru-lumen of the Fogarty and then the green represents the segmented mass effect balloon. Here we see a pressure and volt response. This is in orange our ICP and in blue our impedance. Both are increasing as we inflate this balloon over time. This is raw data for one element for one pig. If we then consider all of our elements and within a pig our ability to detect this change as we inflate our balloon, that's what we see in this graph here. The red bar represents our threshold for detection which was determined on a benchtop setup based on our system's SNR. When we consider across all pigs, this is for all nine pigs. While there is variability across the response as you would expect and is commonly seen in animal models, all nine out of nine detected. In fact the average detection volume across all pigs was 0.39 milliliters or just under 400 microliters. When we consider these two responses here on the left, once again we have that raw data for our pressure and impedance within one element for balloon inflation, but what happens when we look at the same element in the same pig during blood injection? Pressure is still increasing but our impedance trace, that blue trace, interestingly is now going negative. This is really, really interesting. If we actually look at the correlations between these two, we see that on the left our inflation has a correlation coefficient of 0.946 and on the right, our blood injection, a correlation of negative 0.493. This shows that impedance, while responding to pressure and tracking well, is also able to be sensitive to that volume component and this is essential for our potential to be able to differentiate focal from global and ischemic from hemorrhagic. Here we see, if we then take those raw data and sum across all of our pigs, that if we have our pressure alone, it was successfully able to detect an intracranial volume event from baseline. This is great, this is what it's supposed to do, this is what we would expect since it's used clinically. In there, in our intracranial event, we have both a blood injection and a balloon inflation. If we then look at this split out by pressure, while pressure did detect the blood and inflation change from baseline, it did not find any significant difference between them. Here that range on blood is slightly smaller because as we found out, our volume did not stay as focal or localized to the brain in our blood injection as it did in inflation. Even such, pressure still did not detect between them. If we consider impedance as an additional monitoring modality, here we see that the change in impedance during inflation and hematoma, during those same two events in equal volumes, was significantly different and able to differentiate which event type was occurring. This is really exciting because it's shown that not only, like the previous result, are we detecting, hey, there's an intracranial event happening at all, but what type underlying etiology is behind this event. Looking forward, we are excited to pursue potential non-invasive applications for this monitoring. Specifically, if we remove this central electrode and only look at peripheral electrodes, what happens to our sensitivity? Are there strategies we can have that might help overcome this high impedance barrier of the skull? We've actually already started on this work and are seeing very exciting results. Further, we have data with all the compliance balloon that was out of the scope for this talk. We're just starting to dive into that now and we have the potential using impedance to not only detect volume instead of pressure or in addition to pressure, but compliance as well. And then lastly, we're excited to continue to collect data and further investigate this focal versus global differentiation, as well as potential localization, as this color map right here shows we may be able to do in the intracranial space. We validated this in phantoms. I would like to acknowledge all of my co-authors, as well as the key technical support without whom none of this would be possible, our funding sources, the NIH, NHIRC, and NIBIB, who support me through a training grant, as well as everyone in our lab out at Dartmouth. Thank you.

electrical impedance

secondary injury

real-time monitoring

bioimpedance pig model

traumatic brain injury (TBI)