 Okay, so as the very last lecture of the summer school, it's my honor to introduce Oscar Camara. He's the leader of our research group and the person I work close together with. And while I'm working more clinical, he's much more into the simulation part. And so he's currently doing some very, very interesting work on looking at chicken wings and cacti and wind socks and things like that. So, Oscar, please. Thank you, Art. And thank you to the organizers for inviting me to talk again this year. I guess that putting me the last one is because they know I'm a clown, a stupid clown, and I wouldn't put a lot of equations and just call it for movies and pictures. So thank you again, organizers. So just following this, well, I challenge just to put a bit of weird title like chicken wings, cacti, wind socks, and cauliflower in cardiology. So that's really a weird question. I mean, some of you for sure already know what is this about. But some, maybe you are in the position as Aurelio when Aurelio organizing the talks, et cetera. He asked me about the title. I send him this, and then he was like, sorry. Pregunta tonta, but is one of your jokes or is this serious? So it's serious. So, well, I mean, and I send him the abstract a bit later and maybe he understood a bit what was going on. The thing is, last year I needed to talk about electrophysiological modeling. I was super happy about it also. It was thanks to Bart and we had a past conversation and then after the VPH, the first VPH summer school we got another conversation if it worked. So Bart was very happy, was very successful. The first edition, I was barely just remembering the beer tasting and that I needed to give a one hour talk on EP modeling, forced by him. And then, well, he was telling that this year will be again a new summer school and on flow. And I was thinking about Carlos and Andy with their flow thingies, nothing related to work. And Bart asked me, can you give a talk? And I was like, come on, Bart. I mean, I didn't get the ERC grant, the Consolidator grant, because I wasn't an expert on flow and I have your stock. So, and then Bart is asking me to talk about flow. And he was like, well, but you can talk about chicken wings. And I was like, okay, we have the beer tasting and now we talk about chicken wings. So the gastronomy summer school in Barcelona. But then he convinced me saying that I could escape from my paternity leave for several hours. So then I accept it immediately. And that's why I'm here. But again, I mean, my relation with Navier Stokes, these two guys, French and UK guys, I mean, has not been very intense over the years. I mean, when I started doing a bit of research, I was more on image processing and of oncological images or modeling of Alzheimer's disease. Nothing related really with flow. Then when I arrived to UPF, I switched a bit to the heart through all the years of my Ramonica Hall Fellowship and then being an associate professor in our group, FISANS, doing a lot of personalized cardiac modeling, VPH, and more on electrophysiology and a bit of mechanics, but also mainly just going for data processing and integration. And I mean, being in a large group, in large groups of researchers around cardiovascular applications, obviously I was collaborating in some flow related works like these two ones that I will briefly explain now. So the first one was with Simone Baloco, a postdoc that was at the moment at CISTIP, where we tried to estimate the mechanical properties of cerebral aneurysms from imaging. So we embedded or we developed a whole data simulation framework where in the bio-mechanical part, we were using FSI to impose some constraints on the simulation on the structural simulation. So we had some simple simulations, flow simulations on with console and some flow results, flow patterns in some cerebral aneurysms, but not more than that. Then, oops, here, there's something missing, okay. I haven't put it here. Well, there was another thing with blood flow in the left ventricle that I collaborate with Ali Pashai, another postdoc that was around, but I was mainly working on how to compare these fluid simulations in the left ventricle with phase contrast MRI data in very few cases. So then, in 2015, we were starting to think, okay, that this should apply to this ERC consolidated grant. And then, I mean, my background really, I was working, I work on image processing, on modeling, on neurology and cardiology. So one good point was to go towards an application and methodologies combining and integrating the neurological and the cardiological part. Both in application and in the methodological side. Because there are some medical problems where really affect both the brain and the heart, and it would be very interesting to have a more systemic and holistic approach towards this, to study these diseases. So two examples came to our minds doing the brainstorming and one was just talking or thinking about all people and it was the relation between stroke and actual fibrillation, something that is well known, very well known. And that's why this was not innovative enough, not high risk, high gain, enough for an ERC. So then we went to the second idea, more dealing with young ones, and it was just to study the influence of cardiac abnormalities in brain development. So as I was saying, nowadays in hospitals, really in most of medical problems, they go to the heart wing or the brain wing or the neuro wing. And there is a lack of this kind of more systems medicine approach that is improving. So that's why this was the title of the project, the Heart and Brain Axis, and just how to link these problems, the cardiovascular level towards the brain development. So we thought about the whole, we developed a whole idea about to work on data from intrauterine growth restriction cases where it has been already proven that they have cardiovascular remodeling and they have some strange changes in the brain development that they are not well explained with the current theories of mechanical brain development. So the idea was just to prove that the component of the hemodynamic and vascular changes was missing in this equation. So we developed, obviously it involved a lot of different data, human, experimental, data processing at the different levels, multi-scale, and also coupling a mechanical part with lab models and with flow related simulations. Well, luckily the first year they liked a lot the idea, the reviewers, I was invited to Brussels and after an intense session, the 12 reviewers decided that they weren't going to fund the project in the end. The following year we improved a lot the project and the reviewers thought that it was a very bad idea, the same project. So, well, I mean, this project hasn't started yet, but it will, so if you have any ideas or funding just to contribute, just very happy, but we'll start this quite soon. So we say, okay, let's go for the second idea, the relation between stroke and natural fibrillation. Here we can see how, well, it was very quick, but this relation that is very well known is when you have natural fibrillation that is the abnormal rhythm in the heart, so the blood can clot in particular in the atria and then go up and end up in the vasculature of the brain and then induce an stroke. So this is quite critical and it's well known, but the tools, the knowledge in here we'll see during the presentation that there are a lot of room for improvement and a lot of questions to be solved yet. So in particular some clinical data, it is around one-third, around one-third or 25% or one-fourth of strokes, overall strokes and bodily strokes are related to atrial fibrillation, so it's quite massive. So as I was saying, in the atria, due to blood abnormalities, the rhythm of the atria, some thrombus are created and then can go up to the brain. And something quite interesting is just that this thrombus formation is not very well known, why it happens, when it happens. It's a very difficult problem to study also. There are some hypotheses in terms of the characteristics of the properties of the endothelial wall, how they get degenerated, the relation between the velocity, the flow velocity, the blood flow velocity, how it's related to the thrombus formation. And one very interesting number is that 90% of all the stroke-related thrombus are formed in the left atrial appendage. Then not a lot of people know. In fact, there are some papers from the 2000s say that this is the most lethal appendage we have in the body. So this is some pictures from Professor Damien Sanchez Quintana, collaborator and group friend in Extremadura, one of the most well-known anatomies of the heart in Europe. That he has a lot of post-mortem specimens and here are some of them. So this is the left atrial appendage, open inside just a close, well, attached to the left atrial. Here is from outside. This is the aurifice. And we can see it here in another different way. Here the left ventricle. So it's a very complex structure. And in fact, it's involved in several pathologies like atrial fibrillation. And it's not very well known why it's there. And this is quite shocking. I mean, because after all the knowledge we have in anatomy, et cetera, nobody is completely sure why in the hell evolution left these appendix there. There are some hypothesis, obviously, like a possible decompression chamber role when in the atrial there are high pressures like in leventicle systole or in atrial fibrillation. But this is yet to be proven and demonstrated. Also, some very interesting characteristic of the left atrial appendage is that they are very different in people. So there's a high interpatient anatomical variability. You can have different lengths, different number of lobes here to some numbers. This is the different number of lobes or if it's dimensions, the overall shape. And we don't know why. Why evolution left this appendage there and they allowed also to have a lot of different shapes. So this is a question yet to be answered. So that's very interesting. Since long ago, there were some people trying to study this structure, but it wasn't very easy to get data. Anatomists, there were some studies already in the late 90s just from autopsy specimens, no imaging, no medical imaging there. But it's very funny because they look like the pioneers of 3D printing. They were using these molds of synthetic resin to just build these kind of molds and then measure these molds. So you can see that there is a large range of volumes of orifice diameters of length in all this population. Also they saw, it's a very difficult structure to measure in an objective way. How you measure the length in here in a very curved structure, where do you measure the width of this structure is not obvious. And also they found out that the diameters were increasing with age, also quite interesting. And then some people with the advent of medical imaging, they were starting to look the geometry or the anatomy, the shape of this, after a appendage with more detail and then they started to use whatever name to describe these anatomists. So in here, this group, they started to call these lefathrial appendages like having a whole shoe form or a hand finger or fan or wing hook, wedge, swan, depending on the location and orientation of the lefathrial appendage tip, the apex, the distal part. And then they also found out that some lefathrial appendages with a particular shape were more prone to have thrombus. So that was quite interesting. So thrombus formation was related in some way to the shape of the lefathrial appendage. And then here we discover why we talk about chicken wings and whatever. Because in some more recent papers they have started using these kind of categories to define the shapes of some lefathrial appendages and the clinical community they like. They like it a lot. People working on this. And then now everyone is using this kind of classification. This classification is based on looking at the lobes. If you have a large or dominant lobe, depends also in the number of bends and the number of secondary lobes on the angles between the main lobe, the osteum, and the tip of the apex. So then you can have some cactus in here where you have secondary lobes. You can have a chicken wing that is quite smooth. It's quite long. It's quite big. And not a lot of secondary lobes. You can have this windsock that it has a variant. It's a more complex one. They call it cauliflower for whatever reason. Don't ask me why. So I hope that now Aurelio and the likes of Aurelio know a bit more about these cauliflower in cardiology. But this is how they define or they classify the shape of lefathrial appendage nowadays. I guess you can see that it's not a very rigorous and objective way of classifying shapes. And we'll see that later. Also very important, not just shape is flow in order to study the relation with thrombus formation. And in fact some papers have found some link between low velocities and low blood flow velocities in the lefathrial appendage link with a high risk of stroke or even the presence of thrombus. These measurements are taken basically from transesophageal echogardiography. Just measure at one point in the ostium of the lefathrial appendage. Also some they found that the decreased velocities in the aetria are also related to thrombus formation. And also very interestingly, as I was saying before some shapes like non-chicken wing morphologies they are linked to be safe in terms of thrombus formation. So if you have a chicken wing, if you have a chicken wing it's better for you because you won't have less probability to have a thrombus in there. Okay and also some people they found out that the number of lobes are important or related to thrombus formation. So what happens when a patient has a high risk of a stroke? So you have two options. The eternal battle between pharma companies and device companies. Obviously these guys, the drug people are the winners almost always because it's the first option. And in this case it's anticoagulants. The problem like warfaring or the novel oral anticoagulants they look like more performant. The problem is that 20% of patients because of bleeding risk they have contraindications to these oral anticoagulants. But also, I mean this means that these people they need to take a pill for life, every day of their life. So the pharma companies are quite happy about it but people just forget or they don't want to take pills anymore so a lot of this anticoagulant process are not followed and then originate some problems. And some years ago they appear these new devices around 15 or bit more years ago appear these devices, the left actual appendage occluder devices as an alternative to drugs for people with risk of a stroke. So they obtain kind of very, very good results in some registries and some clinical trials. Some of them have already been approved by the FDA like the Watchmen and there is a whole community working around these devices and several companies and it's getting more and more market because they are seen as an alternative for patients not just the ones that have contraindications to oral anticoagulants but also as an alternative. So the first one were like this one like this that was called Plateau very simple that is not anymore in the market and then this is the Watchmen one of the most used and this is the two from Boston Scientific and these are two versions of the Amplatser the San Jude one also very, very much used and the idea is just to put in here this left actual appendage to occlude it and then just trombosis cannot be formed anymore in there it's just really to occlude and to cover the left actual appendage. What happens in the long term? No one knows. No one knows. There are no studies. Long term longitudinal follow up studies for these devices. So now they are becoming to appear some studies the five year follow up for the first one or some of the first one presented so we don't know and this is yet another question to answer what it happens in terms of long term you are covering one part of your atria and yeah probably there will be some consequences. Some anatomists they started just to look carefully at some of the orifice dimensions in order to better know which devices were better to be implanted because I haven't mentioned that these devices come in different shapes because left of appendage have different shapes in every patient and this is the most important decision clinicians need to take when implanting this device is the size of the device because it needs to be small enough to fit but large enough that it doesn't embolize it doesn't get out once it's introduced and this is the clinical question the clinical decision about these devices besides using one from one company or another this sometimes is related to other things. Okay and then obviously you cannot have the post-mortem heart in order to measure now what they are doing is to use imaging imaging techniques and they use several ones basically they use X-ray and TEEEco transphysiological echo also during the intervention they do it before the intervention for planning and also during the intervention to check and in some clinical centers they are starting to have also some CTs in order to have high resolution information of the anatomy but most of them they are just using these two and the problem is that they get contradiction information everyone working with X-ray or echo they know that they are very noisy type of medical images and what they do is just they do some measurements and then based on these measurements they decide the size of the device to be implanted so let me see if it works the web just to show you on the web you have some live cases this is Professor Horst Sieber from Frankfurt I will put the boys I went in November to one workshop where 300 the most important people and clinicians working on this type of devices worldwide were there they meet several times per year and they do live cases and this is quite spectacular the guy did in one day I don't know seven or eight of these interventions live and with a micro and just explaining the procedure and it was quite funny so they present the case whatever and the guy was having help from other clinicians worldwide that were helping on the intervention and what it was funny at least to me also was that the clinical decision was made by all audience so I mean this patient was treated by 300 clinicians the world experts on this so probably he wasn't happy about it and they were discussing well would you implant a 27 millimeter watchman and people were voting okay let's go there it was quite surprising some clinicians told me that in order to be able to operate and talk and do a presentation at the same time you need to get used because it can be quite messy so you can see that they check also the orifice of the lefathal appendage with ECHO 3D ECHO also this is an echocardiographer from UK that was there just helping he was in the conference and you see this is the appendage if you are not an expert or you are not used to this it's challenging and it's moving obviously it's moving so they take the measures and this is a 2D view of the probe the 3D probe of the ECHO probe that is moving also so taking measurements here is not obvious and it can be more objective they also check flow with Doppler obviously and let's advance so people were discussing there so you can see actually there isn't much lobby on that point so they are talking about the number of lows where they will implant what they call the landing zone how deep they will implant the device or not and they were just discussing laughing and so they were having a lot of fun and to get the right views of the ECHO just to get the proper measurements is not obvious at all so there is a lot of room for improvement in terms of imaging and then you come to this the magical x-ray that only understand clinicians because normal people don't see anything in here but gray shades but they use this just to guide the wires they introduce and they go transceptile through the atrium so they make a hole in the septal atrium just to reach the left atrium and just let me show you so this is the wires that obviously is quite clear that it's already in the left atrium and then they take some measurements super robust based on these shades of gray or whatever and yeah obviously 25 22 millimeters 22 not 21 and then they say it's a pity that it's a bit inconsistent what we measured in the ECHO what a surprise and then they do okay in ECHO we see 23 in x-ray we see 27 let's put a 29 come on guys and this is how they decide the size really and what is working quite well I mean it's 95% of the devices are well implanted according to their criteria so why you want to complicate things but it would be interesting at least so this is I'm not going to go through all the implantation but here the device that is magical devices I mean in terms of material properties is something that you put into a catheter and then expands there just to me it's a bit magical how this material works and then this is inflated and it covers what it needs to cover and this is the implant being already put in the lefatele appendage they do a bit of what they call is tack or tick tack whatever just to see if the device is not going to get out easily one is implanted and then they put some contrast to check if there is more flow or not entering the lefatele appendage let's see if I can get some of the you see this is kind of one of the umbrellas of the device that it's already put in there and they do more measurements just to check if everything if you can see more or less the flow doesn't enter the lefatele appendage anymore because it's covered and then they do whatever more echo and you see that the lefatele appendage cannot be seen almost anymore this is the device and it's covered so this is how the procedure is and this is one of the devices oops so I mean being an engineer and a computational guy I think is quite interesting because there are a lot of interesting things to do in here one of of the things is just improving the imaging so there are new imaging modalities like this 4D flow MR and some papers already showing that you can have blood flow not just with 3D echo but with 4D MR in order to study blood flow in there in 4D, proper 4D with esthesis maps etc also modeling that's quite interesting and there are very very few attempts of CFD because it's very difficult to get proper geometries of this lefatele appendage and the papers nowadays is just done on one, two cases or even synthetic and no link between hemodynamics and anatomy so when you see that as I was saying I was lucky enough to be part of a very large group with a lot of talented PhD students working at least three or four of them on cerebral aneurysm it was part of a big European project called aneurysm and there were people working on simulations on CFD simulations, on device simulations on image processing of cerebral aneurysms and when you see the pipeline of the computational pipeline of cerebral aneurysms and you see what is needed for lefatele appendage it's exactly the same it's exactly the same so that was quite useful to realize also I got involved with the hemodynamic unit at hospital clinic because of some 3D printing projects doing some funny thing with for devices or systems for electrophysiologies training or for TABI planning so all this together will rise into compilau that is my current national project for three years where the main objective is just to develop computational tools to study the 3D morphology of the lefatele appendage and the hemodynamics before and after this type of law interventions and it's very similar to cerebral aneurysms things that we want to try is just to find ways to build statistical analysis to relate shape with hemodynamic parameters how this shape changes hemodynamics what is the role in the whole cardiovascular system of the lefatele appendage what happens when you close when you clear the lefatele appendage which parameters are the more important related to thrombus formation all these are not answered and the jackpot is like can we have a model to predict if a patient needs anticoagulants or just go directly with a kind of device I mean if you get that just you will have a lot of funding from a pharma companies and then I mean just this is just more specific objectives I won't go into detail but it involves data processing kind of simulations so we are going to use post-mortem data and 3D printing to generate ground truth data to validate the simulations everything will be open access for everyone to play with and try to translate some of these tools in the clinic so I'm going to show you a bit of the preliminary results we have I mean it's collaboration with the hospital clinic the hemodynamic unit and some people in Belgium some radiologists there quite interesting because in Barcelona they use the amplator device and in Belgium they use the watch hand so basically typical acquisition studies with TE and to the x-ray but also in Belgium we have 3D and geography data that's quite useful to obtain 3D data and the geometries and also we'll play with post-mortem data and Andrew Cook in London what else? yeah I mean this is just the acquisition protocol I won't go also we are in contact with Materialize the big 3D printing company just for this development of 3D models and then connect them to a pump in order to generate ground truth data also with people like Hernan and Matthew in Philips just on this side of validation of flow simulations and this is some of the data we start to have around 40 geometric some more complicated in terms of having pressure information also before and after the intervention some with good and complete clinical studies other just regular actual fibrillation cases and we are starting to play with it so these are some of these left actual appendages we have in our database just challenging just challenging all of these are left actual appendages we had some undergraduate students fighting with them trying to follow the standard criteria to separate and to classify them into this kind of cactus chicken wing etc so we have observer dependent classification of this it's okay having checked by clinicians and it looks okay because one day it can look a chicken wing and the following day for the same person can look whatever a cauliflower we have started to do simple measurements on these cases like diameters, height so briefly some fractal analysis studies to check but not very promising in order just at least to have some dimensions of the length, the width, the diameters and relate with simulations so this is the whole pipeline that Andy and the Oliveris around there is one of the if not the most important person in the project together with Lupe, a student of our master just developing this pipeline for generating finite element models out of these geometries it took us a lot of time to get something more or less robust we also tried mimics from materialize but we thought that with the current pipeline we have a lot cheaper all this is open source and free and well it takes a bit more time but with faith and care and patience you can get similar results it involves as you can say from the segmentation just to generate surface meshes, some smoothing some manual editing for the pulmonary veins mitral valve area determination more post processing of the meshes until you reach the final 3D volume mesh and then running CFD simulations until now with ANSIES but also we are going to use Alia from the BSC etc and the post processing with Parabu so let me show you some real nice examples that Guadalupe generated on some cases this is what we have from the rotational geography images from Belgium so segmentation is very basic from region growing techniques semi-automatic by people in there then marching cubes and then we need to just play with mesh lab etc just to remove this catheter that appears there we need just to do some smoothing to run the simulations and Tobin smoothing is good for volume preservation we need to just determine the pulmonary veins and the mitral valve also for the boundary conditions what else and more whatever operations just to just be able to put the boundary conditions and the pulmonary veins in there just put these cylinders and then the tetralalamation so we need in some cases post processing of the mesh just to manually change and separate from the left atrial appendage etc but it's working not sure how many cases we have already finished but more than 10, 15 so something like that and obviously we have a lot of we are starting to collect a lot of different parameters from the hemodynamic simulations also we have kind of a very ideal oval atria where we have plugged a lot of different appendages in order just to check the influence of just changing the appendage morphology so this is quite useful just to be independent of the shape of the atria so we have some boundary conditions that we have found in the literature in terms of pressures, velocities and the mitral valve and the pulmonary veins blood is modelled as newtonian incompressible fluid something more or less standard the mitral valve is treated as a wall at systole and then open at diastole this is something we need to improve to have a more continuous the whole cycle thing and these are some of the first initial simulations here we have the left atrial appendage and some streamlines that no one can understand but clinicians like a lot colourful things so we showed to this thumb depoter and immediately he showed that to the people the clinicians in Frankfurt he didn't know anything about what it was but he liked a lot the colours so he showed and people wow so it was quite interesting these are more colourful pictures that we are in the process of trying to understand with the oval atria we can see in systole how it enters there in the left atrial appendage but not that much in diastole this is kind of the mitral valve mitral valve velocities around the whole area so we are starting to get some of these parameters to analyse them some of the initial interesting results is that this is just all the different appendages plugged into the oval to the oval atria and I mean the velocity profile look very similar independently of independently of the left atrial appendage shape so that's something for the mitral valve velocities it looks like the shape of the left atrial appendage is not important even we remove it like it was an occlusion it's around there it doesn't look very important on the other hand when you check the velocities within the left atrial appendage there are differences between the different morphologies that's clear and we are now in the process of analysing this how we analyse? so typical measurements from the hemodynamics looking at wall shear stresses looking at because in theory low values of wall shear stresses related to low velocities related to more risk of thrombus formation oscillatory shear index also more related to the complexity of the flow if it's more complex it seems or it should reflect more complex geometries and also more prone to have thrombus formation resident times and also one very interesting that we recently discovered from Humphreys and Alberto Figueroa on aortic aneurysms is this endothelial cell activation potential where it combines the OSI and the wall shear stress meaning that if you have very complex patterns a high OSI and you have low velocities low wall shear stress then you have more risk of thrombus formation so that's quite an interesting measure so we are starting to have tables like this one full of parameters for morphology full of parameters of hemodynamics also videos here like vorticity and we are starting to suffer from big data sickness we don't know what the hell to do with all this data very well so we are really nowadays in the process of trying to make sense of all this we also have with the real the realistic geometries this is just for cases and we are starting obviously colorful and nice pictures on vorticity renal maps streamlines residence times but obviously this is completely useless if we don't try to quantify this and try to relate it to some clinical hypotheses also this is wall shear stress maps and the vorticity values I mean the initial things is that they look okay so vorticities are created in the osteum of the left lateral appendage and not on the other parts of the left atria with more complex geometries you see more vortices etc higher values of stress this is the values of this cap this relation between the ocean and the wall stress and you see that the red parts below are the regions with higher a cap value meaning more risk of thermal formation and it's logical that appear dear and in this type of configuration also full of values for the realistic geometries and this is just for cases we have 40 so well we'll see what we do with the rest so yeah some of the initial conclusions that we have until now it's just that there are a lot of differences in volumes left atrial volumes pulmonary vein orientation and shapes left atrial appendage volumes landing zones lengths and number of blobs nobody has taken this into account to check for all these parameters influence on hemodynamics also this mitral valve velocity is independent of the shape it's interesting also that using a single point measurement to evaluate the complexity of a flow pattern it's just stupid they measure just at one point on the TE images and they say okay we have the empty the historic blood flow and that's all it's like in electrophysiology when you use the total activation time to characterize an electrical pattern it's too simple you need better than this and other things like yeah I mean the vortices are related it looks are related with geometries with more lobes and that this chicken wing cauliflower is just too subjective it's not very useful shape analysis when we look at this we start to be a bit desperate because we are trying a lot of things and people like smart people like mathematician people like Costa etc and we are starting to play and we would like to have to generate a space a common reference space where you can compare these things but it's not simple at all and and to compute distances find similarities we try some spectral methods where we find some correspondences etc and whatever dirty word in maths Costa has tried auto diffusion, heat diffusion placebo trimmick coefficients whatever we are playing really and we don't know what we haven't found anything really useful yet we have some undergraduate students that we are slavering a bit but they are so talented that in three months they are doing fantastic things and we ask them just to try yeah I mean just do cut models of these devices find them on internet on the patents or whatever and they did it and that's amazing that's brilliant and in fact we are starting just to virtually implant these devices in some of the geometries and we will start playing around also some other students they are starting to develop together with Patricia García Lam model of the whole cardiovascular system including the lefate and in AF not solved yet but that will help to have more realistic boundary conditions to the fluid simulations and for the work I mean we have these all cases to process to find a more rigorous way to characterize the shape of these appendages one very important limitation we have now is that we treat the appendages as rigid and this is not true at all they can change up to 50% of their volume during the cardiac cycle so we need to add motion, the mitral valve, displacement FSI, etc we will start very soon with an integrated student on practical and this 3D printing part in one year or two start looking into models of thrombus formation so this is the gank of Compilao a lot of people really really nice and really smart and some of the involuntary helpers students here and obviously we are part of a bigger crowd that you know very well the Barcelona medtech and also I need to thank Blau for allowing me to finish this presentation she was quite behaving and that's all thank you for your attention if you have any questions thanks a lot for this great presentation very nice as always entertaining as always and I think what's also very important is that you keep on talking about the fact that this pipeline is really important and when you do modeling and you say we're gonna model flow it's not so that you just model flow it's like it's a whole pipeline and also like maybe you can very very briefly share your experiences like how do you set up this pipeline because one of the things is obviously that you need to implement is going finding the tools and things like that but also you need to get all these people engaged because you need to get your data somewhere instead of just using some kind of arbitrary geometry so how do you get on with this is this something that you say I start my project and one week later you have a simulation or how should people try to approach this it's quite interesting because the modeling problem, VPH problems the pipelines could have kind of similar faces you could find similar faces whatever you are talking about mechanical modeling or flow modeling or electrophysiology but depending on the question and on the physical phenomena you will have more problems with different of these steps and in fact in here when you start working with flow something that is quite different from other physical phenomena is that you don't worry too much about the equations it's navier stokes, that's all you don't need to go and invent whatever, just forget I mean it's just navier stokes you will use navier stokes and if you pay or even whatever you run it's a button you can run it and that's all you don't need to code so you need to worry just on generating the meshes and generating or have the appropriate boundary conditions when you go for mechanical modeling of the heart you really need to worry on the equations also and electrophysiology also you have different models and you don't know what to use but it doesn't mean that this is easy what you were saying is the whole pipeline and when you see the reduced amount of scientific works doing modeling on this particular application I mean there are some difficulties and one is just to have right geometries and right models and the thing is we had good partners and good friends working for years on this cerebral aneurysms or other type of modeling problems where you can ask them and they will tell you perfectly try to use this measure try to use G-Mesh try to use MeshLab or MeshMixer because it will help you for manual editing etc but still nothing automatic it involves always manual and painful addition so you need just to get the data from clinicians have people with experience on modeling both on the machine side and running equations etc and then have some validation data and just try to put together everyone everyone is interested so networking is as important as the modeling itself and maybe I cannot let you get away with it obviously since you mentioned the big data deluge what about using machine learning I tried to provoke a bit on this what you would do here you put all this data part simulated part not simulated not validated at all deep learning and what and it will say this parameter is the most important well done and now maybe we will end up doing it just for fun because it's not that difficult to do this is really the black box but we want to go a bit beyond that and try to understand why these parameters and which ones in theory they should be important or not we'll probably start using some machine learning techniques to try to make a bit to have help I mean machine learning is very good to have help and more insight on the data not just to treat it as a black box other questions thank you for the very interesting talk functional mitral regurgitation is a huge unmet clinical need and expected to be a bigger problem than aortic stenosis in your model how will you introduce that as an outlet variable just to see the impact of that just because I think that might drive clinical change and we may treat functional mitral regurgitation you can prove it has an impact on the stroke risk and left atrium it would be super interesting but I think we are a bit far from it I mean obviously it will involve changing the boundary conditions at the mitral valve and that there are some people we know that we've been working with I mean Phillips etc working on mitral valve modelling so one possibility would be try to couple our model with their models probably in a week way for starting and try to use their output as boundary conditions input boundary conditions for our models but I think we have a lot of problems to solve before going to have detailed mitral valve situation but for sure it will have an influence I mean obviously so she will increase the pressures of the atria and so question there thanks Oka for their nice presentation couple of questions which modality is used for the reconstruction of the atria this is extra 3D rotational and geography and how long does it take to process one case to have it ready to run segmentation is done I mean the machine is general electric sorry they have tools for doing easily the segmentation and then I mean now we have run the pipeline for several cases Andy what couple of days just when you have you have your raw data your image and then you have okay I'm ready to click the machine the volumetric mesh okay and then could you comment on your results very curious why in the velocity at the entrance when you have the same idolized atria and you have at the entrance of the appendage you have the same velocity no matter the shape of the of the appendage could you comment on that iPhone at the mitral valve not at the entrance okay and that's something we need to see why why it looks like is not it looks that it doesn't change the mitral valve velocity profile at the entrance of the left atrial appendage it will be different for sure because you have different different dimensions different blood flow entering I mean we have the same pulmonary vein boundary conditions but the left atrial appendage geometry is different so there they are different okay makes sense well then maybe the answer for that is that the influence of the appendage is you can neglect that with respect to the whole hemodynamic in the in the cavity well the first thing is we need to check when we add movement meaning more realistic this is still valid because right now we could say okay now we are modeling just the worst case scenario of persistent atrial fibrillation so you don't have a lot of movement it's an excuse to say that we have an added motion but in this case of a severe atrial fibrillation the atrial is not going to move a lot but we need to check if when we add a bit of motion it's still the same or not yeah I was going to ask if apart from this subject in morphology descriptions did you try to use the measurements of the flow that you are collecting to classify or propose a new classification of those shapes we are in the process of trying to understand them or you mean classify shapes based on the hemodynamics yes and try to identify which is for example the most discriminative parameter this is the idea of this joint analysis what we want to try is just to analyze jointly morpho and hemodynamic because we think it makes sense can suggest to use random forest or this random forest which is not deep learning no no I mean we have a lot of people working for years and years like Cecilia with decision trees and any machine learning kind of high machine learning will say decision trees is just too simple whatever but we believe that this is interpretable so with random forest things like that we like a lot also if I missed it and I'm not a cardiovascular person so naive question maybe does everybody have one of these or appendage yes and do we know when it forms is it something that happens yes I have some slides from the anatomies of Extremadura and I try to understand them every time I see them again but Bart can correct me he is the expert on everything related to heart embryo logically it's a different structure from the left atrium in fact it starts to be formed before the left atrium itself so it's kind of a remnant of the origins of the left atrium that's why you see that it's not a super small surface it has specific muscles and it has some complexity within that is not in the left atrium itself that it develops later on right Bart something like that okay and that's very interesting also that this remain there and with different shapes on everyone so the thing is it's different but it looks like it's okay everyone can live with different shapes so it shouldn't be important the shape for normal functioning but then with abnormalities like thrombus it looks like there is an importance there so that's that's interesting just as a and then I think again it might be something I missed is so how do clinicians decide when to put a device in is there something that's happened nowadays it's when patients they have contraindications to anticoagulants because risk of bleeding but obviously this is just kind of a Taifa kingdom if you're an expert of live atrial appendage devices you want to implant whoever pass in front of you so and they are trying this community to get more important and that means to eat some of the patients the atrial population patients with risk of a stroke but that they are not contraindicated to anticoagulants and this is an interesting fight I mean they are proven no no no we are as good as novel on oral anticoagulants I mean if they prove this that would be quite massive for this community obviously a lot of money I wanted to ask if the left atrial appendage does it have like contravaculation inside? well this is what I was saying they have kind of pectinic muscles that make this kind of column gaudy structure and it's not smooth but in your models you don't model this or no not yet I mean the idea is on the post-mortem data will have from London and from Damien either well whatever maybe synchrotron imaging or microcity to get some cases with highly detailed morphology run some simulations and do exactly what you are doing with the left ventricle and see if we have similar findings good I think we can end here so that also means that this is the end of the summer school with regard to presentations I want to re-tank all of the speakers I think as a program it was very very interesting very varied showing all the aspects that are important for modeling and for what we are doing and so then I would like for you to come back at 2.30 when we will have the presentations of the hands-on and then also where the awards for the posters and the hands-on will be handed out so enjoy your lunch