So I would like to acknowledge an age for founding this work and hopefully I won't forget to acknowledge lots of people it's working at the very end. So today. I'll talk about basically you can divide us into two pieces. The first part is about what you meant to gaming in blood vessels and especially the development of a catheter based device for forward looking imaging. And that technology is based on C. much capacity in which it was an intense thesis. So we'll give a brief description of that how we fabricated which may be interesting for this group as well. And then how we how we integrate CMOS electronics with these devices on on the single silicon chip to build the imaging devices and show some imaging results. Then the in the second part will all talk about something more recent and interesting which is basically. Using thermal mechanical noises India and wire meant to do imaging and sensing using a test to survey like this and as you will see this is really enabled by the fact that you get with the center of the system. You can have a detector. Which has very little low noise. Therefore mechanic in which in a moment can be used as an information source. So the motivation for intervention tool to sound. Is about actually in a way we don't need to talk much about the code of the heart disease being the most significant kind of killer especially in the western world. And over three million angioplasty operations are done per year. Meaning that the. Interventional cardiologist going to the arteries corner arteries they open that up with a balloon or they put a stent in. Those kind of operations and interest cooled us on becomes an important emerging modality in most cases as you see on the left here. If you're using the laser on the left. People use X. ray and geography. To see what's going on in the arteries and for that to see that one day they take a catheter down to a large artery and then they they they pump. This radio pick fluid and that creates a contrast and they can see where the blood is flowing in the arteries. OK. They are going to notice that you see these arteries. There are lots of branches they curl around. So if you want to make a catheter that goes into those places it needs to be pretty flexible and small and thin. For example here you see that that there's a problem there. The look the blood is not flowing. Well. So the the car users would go there. And physically open the inflatable wound to open that artery. So that blocking flow easier and then they missed and put a stent. Metal stent there so that it's stays open afterwards. But what happens. Actually in one nine hundred eighty S. people realize that you may have heart disease. Again and then. But while this is happening. The artery the lumen for the both of you may not get smaller. What happens is that the tissue degenerates and it starts actually inflammation starts building up on the outside of the artery. And you don't see that information by looking at where the blood is flowing. So you need to you need to get information about the vessel wall structure mechanical structure. And after a while. Only if it gets really worse you see the. The narrowing of the of the blood vessel and in some cases these this structure which is live it and the fat in it that can rupture and give you a heart attack and like two months ago on C.B.S. there was there was a program which which was showing especially for women. They actually have this guy. No problems and it gets undetected using angiography they go in and they have a heart attack or two days later they have the injury. OK So that becomes an important factor in imaging the vessel as well as where the blood is flowing. So to get that kind of image. What you do is typically if you have this thin catheter. OK And it all goes on to switch it on the side. So it sound waves and then it rotates. So after a while. It's therms indices wait on the other angle so with that you basically get an image like this. OK so what you see here is that in the center black reason is where the blood is flowing. This region. The second region is the. Is the build up in the artery that you see the arterial wall and you see the tissue behind the artery. So if you do some analysis. So that's the blockage in the artery right now. So you can see that easily using an ultrasound because you can get Those are some tissue information which is called virtual histology So in a nutshell what happens is you have a test to see if you turn to the pulse and then. You flip this into receiving more then you see this R.F. data you take the envelope of it and that becomes one line of that image and then you rotates you send a pulse receive. So that's the simplest single transducer rotating system. OK so the current I was catheters that you would find in hospitals. There are two types one is the single element system that we just discussed in this case your this. Catheter an inside this is there's alternating mechanical rotating system with a single transducer I drew Tate's with eight hundred R.P.M. when you take it out. It's that kind of sound. It's a pretty wild. But thing is it is in the cast and there's fluid in there in this between the terraces and inside the cab assailing is pumped all the time. So all those on can transmit and come back. So this gives you high frequency but it is rotation out of it was mechanical rotating it gets stuck sometimes. So the image may not be the best and also it is side looking meaning that in the artery you only get information on the cross-section. It doesn't see for the other one which is getting more traction is using an array in this case what we have is a direct link in the front and then behind that and it's a sixty four elements. And then behind that there is silicon chips of our silicon chips which takes those sixty four elements and then sends out everything in cables you can get sixty four cables out of a one point one millimeter them with a catheter. So it does all the application pulse pulse generation using eight cables. But it's using Piers electrics and they're not very easy to make put in the form of a race because of that the frequency is not that high and the bandwidth is not the broadband. Again it's a side looking system and the company who built the system they tried for five years and spent ten to twenty million dollars to convert it into a whole looking system but they couldn't extract seed because of the manufacturing difficulty even today after fifty years. They cannot guarantee that more than sixty of the sixty four transducer is going to work because it is still partly put together the manual the system so integration is really a key issue. So again this forward looking volumetric image is imaging is an important shortcoming. And if you look at the important problem that this is really relevant is this what's called conical portable collisions where the author is totally blocked. OK. And then usually what the doctors want to do is they want to go in with a catheter then the user guide where in the middle to open that up physically. And then push the cattle forward. Go in and then close that total Okaloosa. But it's very difficult to see what's what you're going on what's going on because you can. See forward in the current images they get busy start here and they can just give it a cross-section not what's going on. Usually what they do is in a branch like this right now they put one guard one on the one artery on one guide wire and I was on the other out there from the side they tried to see what's going on in the other branch. It's very time consuming and reduces the success rate significant. So if you have this well working this happens at least two hundred thousand cases a year so it's not a small fraction and if you're looking system you can not only kind of guy discover interventions you can also got other interventions which are becoming more important like changing the world in the wall in the in the heart and using actually using a very small incision on the outside and using a catheter to push the volves through the heart wall inside the inside the heart itself. Right now it is done by the help of two doctors who just who are there just to provide imaging. From outside one of those samples from the top and the other one goes to the throat from the inside. So the surgeon looks at those two images and tries to put things together. OK. In addition to X. ray. So if you have something that can really be combined with those catheters that can be also very very important applications. So our approach. Actually the one thing that I need to mention is the only system which has some kind of forward looking capabilities a single transducer which gives you actually an image of this cone. But most of the time it's still used it doesn't give you the clue what you had to give it it just gives you this information. It's mostly used for centering the catheter why it is moving forward. So our approach is basically building a catheter where you have this tiny donut in front in front which has both the. Imaging electronics and imaging to surveys and then it's very highly integrated on a single. They can do or not. And then the center is open so that the guard where I can still go. And this guy who I can relate to on an optical fiber and different things. If you want to do a pretty couple occasions to guard that kind of publication and generate a volumetric truly meant to give it in front of the OK So that that's the main application. And if you can combine everything in this donut then this character becomes very very flexible even in the previous case you basically if I go back and look at this catheter this guy basically has nearly a centimeter long section with the piers and when the eye sees. OK So that doesn't go very well with a cardiologist when they want to go through these arteries and be flexible with the tip. So the basic technology to build those kind of devices is cast in like a machine also in sensors. It's basically a diaphragm directed diaphragm with a metal in it. Built in on a rigid substrate it can be like a silicon wafer and what you do is by applying it the C.N.A. signal. You can Y.B. this diaphragm. OK And then that generates the sound and the way back it receives the altos on that vibrates the membrane with a constant voltage you get charged going in and out and you take that current as your Op OK So the key is that this idea is more than one hundred years old. The problem was that you couldn't get really high electric fields in this cavity if you didn't have vacuum sealing so large. I mean in one thousand or one tried this but they did he couldn't get to the high pressure levels because you couldn't evacuate that cavity. And man this is really ideal for this because you need this lateral dimensions to be small because you want that memory very very stiff so that it doesn't collapse and it can resonate at ten twenty megahertz which is what you want. And also you need a small gap here so that you can have this logic to feel OK. So though the theme from technology and photo talk of a used tools provides us with the tools to build these devices and that was the allies around nine hundred ninety five or so and with that. You can as I said you can use a receiver. It has much broader bandwidth. As compared to peers electrics because the diaphragm impedance. Is that very heavily by the fluid. And so the dimensions we have are fifteen to seventeen microns with a couple microns thickness for the membrane and the gap is important we make gaps in order of hundred nanometers. So if you just simply look at the cool surfaces the figure of merit for this device is that. In any actual mechanical device you will see this kind of there's one ratio want to end. OK so that's how much force you get for voltage that you apply and in this point in for the comparison devices. It is basically the electric field the D.C. divided by the Gap times of capacitance that is the figure of matter. OK. Whereas in a peer thirty device it's given by these material properties of coupling coefficient capacitance. And the permittivity and it turns out that if you look at N.. If you go to around two hundred full D.C. bias and one hundred millimeter gap the coupling coefficient all of the transformation becomes the coolant appears electrics. OK now that you're as efficient as a preservative device. If you can go to that kind of dimension. So much machinery enables for us to get there so we have as efficient as this or better than appears electric much easier to fabricate in the small scales and also we have integration capable of. OK So those things come as as advantages. And also the broadband. So how they fabricate these devices. Since we want the most compatible of fabrication process we are using low temperature processes. And the P.C.B. nitrite to run for the big. Centigrade as a member material chromium sacrificially and aluminum electrodes and we recently moved to copper sacrificial layer but those are kind of different materials you can use. And very common tools that you can find american room where we can actually change the lateral dimensions of the of the diaphragm by using by deposing different thicknesses by simple masks. So the process goes like this you start with a silicon substrate which may have a CMOS electronics in it then we put a thick isolation layer and oxide isolation layer and then another isolation in and then the metal the the bottom electrode. Then we put the sacrificial layer. And then we should be deposit more light right through to make the gap. Then we start building. We put the top electrode. OK And then we go along with we build more we cover the top electrode with more might try to get the desired kind of membrane thickness. Then we do the hole as you have seen here on the side and then it's a way to chromium. Then again in under P.C.B. on the vacuum conditions the build up the membrane so that it is taken off and now this is basically sealed around thirty minutes or so and then you will double after afterwards you can shape this thing. As I said if you want small mass in the middle and thin on the sides with other mask. So that's basically the fabrication process even if you do it on CMOS we follow more or less the same exact recipe. So again did the motivation for integration. Should be clear from this example so on the upper right you see the conventional doors on machine. You know the proof of the probe which is used outside of the body. That's where it contacts the body and then there's now a twenty six years or the cables carry. All the sixty four hundred twenty eight signals and then this box gets them amplifies them and then do the be informing and then put the image on the screen. But if you want to go to catheters you have to reduce that cable count. You have to reduce the palace's capacitances because as you'll see later these elements become only seventy by seventy microns and they have only two hundred come too far out of capacitance so you can reduce the cable there you have to actually provide right at the source. So you can drive the cable. And also into some multiplex in because you need to reduce the cable count around eight or ten. We don't get improved sensitivity as well. And the challenge is you want to get the bandwidth. You don't have much area as I said maybe want to have millimeters there meters though not so you have and the power cannot exceed hundred fifty million watts because otherwise. Especially when they take the cars that are out and they they leave it to dry the peats up and then start peeling off the materials. So you want to keep your power consumption low as well. So again we actually started this project where we were making the CMOS separately from the CMOS chips and we were doing Y. bonding and but as we said we want to do more integration So right now what I'll talk about is how we call mind. C. much directly on the cmos chip and with that you see the person with capacitance if you look at the wired one pad pad capacitances you get the pick of Out for president as compared to point you pick over that's OK. And then when we do this type of integration we basically remove all that for us to interconnect your presidents and the other thing is for this kind of case you had to do around two hundred Y. ones. OK So in this case you eliminate most of those and then you only down to ten to fifteen connections so reliability and putting things together becomes. And also the cost is going to come down because the menu label is avoided. So this is the this was our first run of devices. So in this case of a what we do is we designed the masks and then sent the commercial CMOS family. And then we received the eighty two a first from them and to get them fabricated here. We basically cut these waivers into two by three inch rectangular pieces so that we can do the processing in M.R.C. and here and. What we get is you get around ten thousand of these devices from from a single wafer that's kind of a problem as well because with the twenty thirty wafers you cover the whole need of of catheters in the world. OK So but we've talked of form families and they would actually support the commercial production because these are very valuable because those catheters are single use and the price is around eight hundred thirty thousand dollars each so you can put some value on these things. And right now the CMOS and man's cost per day will be below to two dollars. OK Again this is. With with our with the university based production and sending getting only twelve waivers from them. So if you increase the numbers that can come down but there's not going to be a significant cost in any case. So this is the first set where we had this even amplify is only now that there is meters and then after the processing we have the chances of being on the same CMOS silicon wafer. That there is because we're external The cemetery was eight hundred forty microns in their meter it had thirty two receivers and twenty four transmitters. Because frequency range was ten to twenty megahertz and to get scared of images from this device we basically what we do is we put this thing going to tip carrier and put give you a petri dish whether we can feel water or the imaging fluid in there. And in this case you see these twenty four transmitters were made on a P.C.B. P.C.B. board an F.P.G.A. controller so that we can transmit things and receive signals. This would be this would for example then be implemented in like this where the transmitters are on the side receiver electronics are here so that would be one step of integration. So with that system we were able to get him three dimensional images in front of the array so this is like three wires put and this is the silicon chip the wires in front of the array and you can see X.-Y. cross sections and the three D. images so that was very encouraging for us and you can get imaging done it. Hundred frames per second rate twenty first or second is real time basically imaging. So you can achieve that easily with this kind of system then we went to the second round of CMOs fabrication and design where we actually put high voltage pulsers also on the same chips and this is this is one of the guys and we share this dive with Jennifer has this group so that they have the one health service share the course. So this can be some CMOS Image or they are an old electronics but we designed the I.C.'s with better performance as full capabilities and we were able to make two point one one point five millimeter raise for different applications this large normal before into cardiac or task catheter valve replacement type and this would be interest killer. For the total cushion image. So these are the devices. Again before and after See my fabrication. Again the pulses go up to twenty five volts and we have basically over one hundred elements here. And one important thing that we have in our design is that we have two rings if you see and wondering this force has mission the other one is for reception and that is very good. That's very nice because we don't need to switch between two has been received and that into the. Noise into the system. Plus you can have the image of a new thought because it's kind of such a fire and the total power is twenty metre watts for the whole thing. What we did was return all of the unused since we only have four channels coming out. We only turn on the four that we use and then turn off and very when we don't use them again most of the power is consumed by the buffers which are driving the cables rather than the policy and this requires only thirteen external connections and this is these are the connections for us right now to probe from the top side. So with these we did initial testing right now we see the only connections we have from the outside world are those those ten thirteen connections against that of that whole big board is now on the chip. And so this is the simulated image of a target. Like five meters away from the array and this is this experimental targets and we use these point targets to get the diversion resolution of the of the imaging was always you know of the array. And the we get good results for simulations from simulations and then this is a more recent results where you want to look at some real things. And in this case this is a chicken heart again placed in front of the array and so if we play this video. Now what we see is that this is in three dimensions you're looking at the cross-section of the heart of think about the heart like this. So we're basically looking at all these core sections so you can easily see. So the area is here and looking up to the heart apex. And you can do three D. rendering so you can see the apex of the heart and the inner and outer wall so again if I if you look at this for example if I just freeze this around here. I can do that. If it works. OK For example they will be here this this could be in the artery for something about Sinatra but this is this is one centimeter with your rays here if there's an opening in the artery you can see the front of the oak Lucian the back of the oak lesion and things like that. And it actually extends so you can going to tissue around seven eight millimeters right now. And so that gives you. And again this is this data is collected at fifty affairs per second rate we're not totally sure how we're going to show this image just of the medical doctors within there will see these images. So that's why just seeing this kind of cross sections. Maybe easier for them. So this is the other cross-section various slicing through like this. So you see the heart wall and then say right around here. You see the wall and there you see the empty blood is a wall in the middle. OK So this is really showing that the system is not capable of seeing through tissue. And we're right now if you also have some data from how twelve's artificial volves just to see if you can see the geometry and then when they're in and they're moving through with our system right now we can make like one or two seconds length movies and we need to buy some Paul who processes to the end of the image we can collect the data but we can have the end of the images fast enough. Again just to compare with we did this year and also highlights our nose performance to get to the next state. So this is a similar development which is going on at Stanford. And in that case they have a similar area but they don't they use a high temperature process so they can not connect their electronics they can come up with direct talks on the same thing. So therefore they have to put those at the electronics on the sides and with that they have to go through T.S.P. so they used to go to the back side and then flex tape and then the circuit. This is basically hundred external connections. Therefore it can not be flexible and also the name is most performance. Again because they don't have a transmitter receiver separation. It's all on fourteen D.B. where there is in our case the noise figure. Meaning that noise is more or less equal and to the tone mechanical noise of the device that's important because in next of I will show you that if you can measures to make an equal movies video system you can actually do something more interesting. OK so if it compares really well in terms of integration I mean this is much more is on a single chip but as you're dealing with mind chips put together with lots of connections. Now just shortly. The two collaborative work going on for cattle integration. So we also made donuts as well. So this is actually through the same device. We have in just about the test these things for imaging and you can use actually flexible interconnect wise to make connections from the top go through this whole sort of backside and you can make make connections afterwards on the backside of this to think our actual cables and that goes along the catheter. OK So that's one way of doing that the other way is. In collaboration with him on the back who is on the back and in this case what the plan is to use the TS piece so that we can make connections from the front to the back side and then use the flex to connection on the back side which much larger area to work with because you have glue and all those other things which we don't want to have happen on the front side. And these are some initial out of the mechanically working devices there's somebody to get activity as well so that those reasons vary had opened up or flex they base that right now there's a couple of T.S.P. is there in the same location. So we have left will regions open mostly most Nordic to the no see much so that you can directly open those things through silicon itself. So this is a very close to getting some results on this. Hopefully in a month or so. So with that I want to kind of jump into my next topic. I told you that with this kind of careful it and integration noise that will come with it really long. OK. Meaning that we may be able to see the noise just coming from the environment around the membranes of the sea much. And also this gives us a simple method of testing our system. Usually when you have a device like this. How do you test its functionality you make you put you connected to terminals you make an impedance measurement network allies but in this case if you look at what's going on. We have this bias on the top and the bottom electrode is there to connect to the electronics. So we don't have access to those two ports. What we have access is the output of the amplifier. So what we want what we can do is if I can measure the noise by just applying a bias and looking at the noise out what I can actually tell if I can see the device information at the output. I can characterize the whole chain not the device the electrons and everything else. The other is we can actually use this for sensing and imaging varies by using the diffuse noise field and imaging work is done with the brain mechanical engineering who uses this type of imaging technique for underwater sonar applications. Again just quickly. We have really low noise amplifiers there Tassie Butin amplifiers. And we can model them and they agreed really well with calculations so we know that the noise performance really well and the important thing is if you go to the analysis the output noise of the amplifier is actually related to the electrical impedance. OK so you have a family with a full K.T.A.R. Johnson noise it's basically if you measure the noise. You know the temperature. You know the resistance. It's very similar to that. Unless you have a good system. Version of the noise at the out. So if I can measure the impedance in a way of the device at the noise. I can get. I can do lots of different things. The first one is again looking at the functionality of the device. So right now if this is the experiment it really nails nothing there just apply a bias. No signals applied. OK. And I'm looking at my it goes through my amplifier and the buffer I'm just looking at the spectral if I look at the noise spectrum with zero bias. I basically get more of the flat line which is expected from the CMOS noise analysis. Mostly much information but when I apply bias to this device which is which then becomes a tester we see that these peaks appear. OK And that's expected why because in a are these are resonant devices. If I had measured the Imperius actively I would get those two peaks and I have two peaks because these devices are actually have two different size membranes so each member in his resonating at a separate frequency and again I can go through the whole array at the output just to test by just using this to control in the system just go to the different elements and I know that all the elements are working OK so you can even use this as a sensor if something happens to the diaphragm mechanical loading chemical composition change these things will shift around so you go resonance sensor already built in without any active loop just to get that noise and then we also want to go in for to detail what happens in immersion because in a of things that are easy there's more caustic activity going on. So for that to be kind of a simulated the system where we do a finite time domain analysis system is not quiet element. So these are different membranes like two different sizes to simulate this and this is symmetry so obviously relating around five elements of the array. This is an immersion. So we can apply forces are actual study forces that are used to membranes and look at the. It's happening all around those other membranes and you have me do that in air we see those two peaks as expected. We saw in the noise because we can actually calculate from this analysis the electrical impedance seen from this element and from that we can go to the noise. OK so we can predict what the noise spectrum should be so again this is video immersion in the immersion we see that at different frequencies. These are the excited membranes for example at low frequencies around here. Only the membranes that you move move around. But when you go around eight point five as one of those peaks we see that only although we only apply the electric field to these guys all the desoldering is moving and that's a problem for imaging because you think that this is only it has been but the whole the reason as many and that's that creates a problem for the imaging now we understand what's going on with this method and from this we can get the thermal noise spectrum expected so moist spectrum and then I mean to the measurement. It is the result in in water again we just put it in a mode of noise sort of similar plight just looking at the noise output zero Also as before. Seventy volts ninety five volts close to collapse. We see that now the C Much is doing something so it's noise is visible on the spectrum and also these peaks are actually coming from two different membranes and they shift in frequency to lower frequencies and increase the bias because because of being so fitting effect that you have in the past of devices. The more bias you apply the more close to collapse you get the softer the side to becomes. And he does. The comparison with the. Prediction. So what we did was we took these curves and subtract the power. That's coming from the noise problem from the electronics to get this. And that's prediction versus theory and they I mean they couldn't be better than this with this complex system and all with all the simplifications that we did. Instead of membranes we use direct angles into the code like to use rigid I mean this linear arrays. And they really well and then you can use for sensing because once we know that things correspond each other for example if I have if I change the speed of sound in the media. OK like five percent below zero or higher. You see that this noise spectrum changes and these peaks move around quite a bit so you can start measuring the few properties by just looking at a single frequency of the noise spectrum so that becomes a simple sensing Macand. Now more interesting way when I started talking with Kareem about this. So he was doing this imaging in underwater applications where the wind the waves and stuff like that they all make noise in the ocean. They all actually sound sources uncorrelated sound sources. So they still travel in the sea and they hit a submarine and they scatter. If you have sensors around by just listening to those and correlating them we can get the image of the submarine. OK So there's a theory behind it and then I'll show you some references about the key idea is the union. Active system you have a transmitter in the receiver you transmit the receiver and then you pulse it and after a certain time delay you get to receive signal. So from the time delay for example you can tell what the speed of sound is what the distance is depending on what what is known in the passive case all these noise phones that are around. They actually they hit this boat both sensors. OK. And if you quoted Late them. It turns out that if you correlate them on the on the negative and positive sides of the time because correlation can be negative and positive time. If you look at for example the negative time. It tells you the information coming from these noise sources which passes sensors to first and then sensor one. And the post of time is wanted to if you look at that you get exactly identical polls as if you're doing possible single you get the same identical signal and that can be shown to Italy. OK so you get to greens function and this actually started very early and people looked at the sun for this now. I mean how it works quickly is that if you have a source like this in the middle that close the correlation peak at the signal at various times of it arrives at the same time on both sensors when when this goes to some other point. It arrives this one earlier and than that one later. So basically by distributed sources you feel this time with all those signals and then if you take the time derivative of this you get these into impulses and that's what you expect from a given function point of view because if you if you pass this. After a certain time to this other signal other sensor at this time or if it's diverse to time it goes to the other side. OK so if you gather all that information and take the time that it would have you get to the correlation function to get a Green function. And if there are many many sources around that basically gives the formation. If you abandon it a system. What happens is only these regions the noise sources around these Asians correlate to this exactly at the same time because for example the noise source here is not going to come to the both senses at the same time. OK. Not around. He did the distance divided by time so it gives the correlated. OK but all of these sources on this line they pass through this guy and that guy with that kind of handling so they end up and then they are given the poles. So now that this is an application of this for underwater sonar where they had an area of these guys and then this is when the new waves from the surface is generating the sound. And they basically had this is actively opened that information. This is. Top of the top of the ocean surface and these are the layers below. And this is by just listening to the noise and correlating them. OK In this case the Iranians are like there's where you have some directivity you're looking downwards with your way so how can you do the simple experiments with our devices at these frequencies. Is that we actually put our device in a water bat and then we called the noise spectrum and we let the water evaporates to the surface came comes down time and we decode the noise and then we can make a comparison. So we have this noise data and then when you go to the correlation. That's what we get. You see time zero correlation. After a while you get one pulse and that's for one point five meters if you get the poles earlier and with a larger amplitude after a certain time or two millimeters at an earlier time you get another part of the pulses in time comes and this will be ideally the case if I had a pulse to see the experiment pulse my signal and it comes back. OK And then if you do that. Frequency time domain analysis we see that in a certain frequency range we get a nice pulse coming in. So this is time. That's frequency. OK So in this we can see a range. I have a really nice and clean pulse echo signal in this because the range here very I have CROSSTALK. I have these events and waves coming in. So there are two types of waves going on here one one of them is waves traveling along the surface which are these the other ones that they go up to the interface and come back. Those are these way. OK we would call again. A lot of anywhere you'd see it on the acoustic where those are the bulk ways. So you would those ones used for. Imaging the other ones you can use for sensing on the surface. So we did imaging with this already. So this is close correlation image on the left. OK for this is a certain. Dept one point three millimeters and distance and this is the regular positive we can use the same rate for both of imaging and we see that we can emit just using the noise at these frequencies and this just came out in the A.P.L. in December of November timeframe. So again just to wrap up for the sake of technology is that we have shown through volumetric imaging capability with these small arrays and we are working on cavity implementation and application of these in other arteries in the in the body is very obvious. You can also have Doppler capability with this because you're looking forward so you can actually look at blood flow variation in the artery which you can not do with side looking devices. Again we can actually combine the loss of space on the tips for different types of sensors that we can put on and we'll be right. Melding felt like a stick imaging experiments because you can have the. It's all the guide what you can have the fiber optic fiber which eliminates the tissue but a short laser pulse and that there is a song which you can detect. So you can actually put the photo acoustic image and the big A little sun is on top of each other one kind of sees the color and it can be very selective depending on the molecule that you're trying to image and the other one just use it down to make of the information so that can be combined easily and the time when they can voice measurements again another important thing about that total mechanical noise measurement is that when you do the pulse experiments you can you have a blind zone because when you're pulsing you can have receive where you have a thermal noise. You're always using very small signals. You can go down to Times you even OK so you can see very close targets and that increases your results in them and the sleep. Another way of doing better imaging is using a venison like surface plausible residences where people use. What we're seeing is a D. acoustic analog of that again. So you can basically have this rape. Lots of membranes in the middle and put a cell in the middle. You can actually move that cell like it'll move everything system and because of the fact that these waves can be very very very slow Like hundred meters meters per second. The ten megahertz that you see of a length of ten microns. You can do even better. So you can get really over twenty five television has been shown already. I did last year was a pure logic alone that we can use those kind of things to move forward to move to some more interesting imaging experiments and that's the acknowledgements call I.C. design we had Jennifer has we had him is processing beamforming expertise we have some biomedical engineer expertise cardiologists for the noise experiments and while not for the TS We work and student post docs from different groups are listed. And this was founded early on board of it's a good foundation that in the middle by Boston Scientific and I age. Thank you very much for your attention. OK so how long how long does it take to basically get a noise base image versus an active image. So that's the catch with the noise imaging. It depends on how much as an hour. You have with noise. So the more you integrate the more your you collected the better lesson are you get but we think we can go down to like one two millisecond range because you can see is pretty high. That gives you a pretty long time for everything. The noise relates the on the side versus the other one. Just depends on your imaging dept as well. For example usually takes twenty. To go back and forth around a centimeter or so. Maybe use a thousand firings thousand times before I received to get the full image so that you see a flame rate of around hundred frames per second. We're trying at least. To start a company right now we're trying to get funding for a company because it's really getting close. There's a company called Matrix Corp something like that we focus on using C. months as resonators for chemical sensing and there were some papers on that but that's that's using the active resonator method. And did the key idea here is there is that you can put many of these things in part of easily so that going to you. Does your noise by simply averaging over space. What are you going to hit is basically reduce the power consumption of those things because we can just look at the noise to get similar information. So that's why we developed that whole model to see what we can do. There are several things you can change the membrane design so that you push the coupling frequencies away from where you want to operate. That's one you can play with the pitch the most common one because that caused the problem is not specific to see much like in Piers a figure a day of the same. What you do is you put a thin layer of mostly material acoustic a lot of material in front of the array. It's like most like so you can rubber that since these waves travel laterally a longer distance. You you kind of damp those waves out and as long as you bring them around say twenty to thirty below. Then the main transmitter. That's acceptable in these devices we didn't have any courting we just put some three microns of properly in isolation. So those are all is well known methods and in in large scale imaging devices you put a lens to do between the piers a click and the human body which is a convex for that you can make easy contact with body that lens for example at ten races across four waves but if you're using it for sensing you may want to get of have that resonance there. So it depends on what you want to do. So those methods. At least one the types that I know are well established and what what they do is they they look at the they get one other line pulse echo and then when when the arteries pressured they take another one and then they look at the correlation. OK so stiff lesions for example they just shift as late. What is this so if you just get they get compressed so and then that's it shows in the correlation between the R.F. lines and from that they can get the stiffness in it. So you can improve that by improving the band with the broader bandwidth you have for the information the better. Those are going to become so we can contribute in that direction but I think. The default or caustic imaging is more promising because there are or there already and that's what we're planning to do for your new Murthy in by engineering. He has certain molecules which a. Change when they when they're in the active most active ocean species which which which is more the most the seen in information areas so that becomes a nice a nice target so you put that molecule it goes to the disease area and you get stuck to it and all of a sudden it becomes of the one of the sort. I'm a woman so that's the that's the kind of approach that we will take on the selective molecule imaging placation and discuss the makes it enables that you can go there and then you can get the sound waves in with high fidelity. Thank you.