Well thanks for the invitation Thank you all for being here again and I misspoke and I'll be talking about using novel strategies for a label free molecular imaging with applications to hit the pathology and particular So when I talk to people and I tell them that I do molecular imaging the vast majority think that I do some type of fluorescent marker when in fact you can actually leverage different types of light matter interactions to extract and dodginess molecular information so this is information that's already in the cells and tissues you don't need to put a new floor force or unexciting the tags in order to extract this information you can use interactions like absorption and scattering which you heard about a little bit with and you also think things like this Persian which measures the changes in refractive index function a way of like these are linear interactions but you can also use non-linear interactions like to photon absorption trancing absorption spectroscopy coherent Raman scattering or even molecular reorientation these interactions require more than one photon to occur but they can interrogate very interesting aspects of cells and tissues including the electronic excited states and their lifetimes vibrational most of the particles or even you can induce the molecule to reorient itself to the electric field so you can use all these interactions to extract molecular information and in this particular talk will be using these types of interactions and the application of his to pathology how can we use the and digest molecular information to help pathologist better diagnose and stage disease. So here's an eye for the talk first go through why has the pathology needs our help and then I'll outline various different strategies to try to extract molecular information from these tissues and how they can help has the pathology so I'll focus and two particular areas one of them will be and melanoma and the other one will be and prostate cancer and also highlight the patients to other types of cancers. So. Melanoma it turns out that it's still a very serious health concern there's about seventy five thousand cases per year and it has the fastest increase in incidence and we're tallied a rate than all other cancers now this curve here looks at the incidence and we're tallied a rate for people older than sixty five years old for melanoma and if one were to analyze this curve and just look at the incidence rate one we very quickly conclude that there must be some type of melanoma epidemic meaning there's a this huge increase in the number of people that are being diagnosed with melanoma can this actually be correct and if so why is this incident were tallied rate not keeping track where the incidence rate what we need two pieces of information one of them is the fact that there's actually not a whole lot of screening and it's very limited and at least during the duration of this graph there were very few changes in treatment now that is change in the past decade or so but for this graph it's true so can there actually be a melanoma epidemic with limited screening and few changes in treatment the answer is No The only reason why the incidence rate can be so high where the mentality rate still increasing but much slower than the incidence rate is because the pathological evaluation does not correlate with the clinical outcomes in other words but just over diagnosing melanoma and that's a huge problem. The problem actually doesn't end there just to say that a patient who actually does have melanoma the next thing that we asked the next thing that we care about is does this patient half met a static disease has the melanoma spread out to other parts of the body and so previously the way that people would assess this is by doing population of the sentinel lymph node basically seeing at the lymph node it's in flame or if it was a little bit agitated it when you applied a little bit of pressure a few years ago that change when they introduce symptom with node biopsy as the primary way of looking at Met a static potential and this study back in twenty fourteen it was shown that by doing a set the length of by. Which is a very invasive procedure you can improve the survival rate five year survival rate by about five percent which is not trivial this is really an important advancement and so now that's become the standard of care but if you look very carefully at the outcome of this study you find some very interesting results in particular there are more people dying of melanoma who had a negative sent them the biopsy then they are with people with a positive sense and with no biopsy. In other words were only able to detect about thirty percent of the patients who end up dying of melanoma for the other seventy percent who died of melanoma it was nothing that the current center of care can do for these particular patients effectively your negative sentiment of biopsy can turn out to be a death sentence because your care stops there and that's a huge problem. Now how do we assess who's a good candidate for a sense of the biopsy it turns out that there's this market called the Breslow thickness that's the primary marker that people use to assess whether or not you should have a symptom of the biopsy but the specificity for that marker is about twelve point five percent mean that for eighty seven point five percent of the patients they derive absolutely no Uncle logic benefit from this very very invasive procedure which has complication rates upwards of about forty percent So again this is not a trivial problem and what we want to do is provide additional markers that can one help us better detect the melanoma and more importantly to better stage the melanoma to see who has an aggressive moment and who doesn't have an aggressive Milena. There are very similar stories in other cancers where the thing that we care about the most is how aggressive is this cancer another type of cancer where this is the case is prostate cancer where the prevalence is extremely high turns out about sixty percent of men over the age of eighty will actually have some form of prostate cancer clearly we're not going to treat all those prostate cancers because not all those not all of those prostate cancer are actually going to kill the patient so what again what we really care about in this particular case is can we just find really aggressive cancers and treat those appropriately. Currently the Gleason score doesn't do a very good job with this and so it is still unclear which patients harbor more aggressive or more idle in form of this type of cancer. Now if you look at many other types of cancers one of the markers that's most alarming is the diagnostic uncertainty meaning how often do pathologist evaluating the saying piece of tissue disagree on the actual diagnosis of that patient and for things like bladder cancer can be upwards of forty percent for melanoma twenty five percent these are huge numbers and they get the problem is that they don't have the right biomarkers the right. Information to make the correct assessment and in fact there was this very interesting study about three years ago where they compare the performance of a pathologist to that of a pigeon and this is true it has five color channels so it's actually very sensitive to the different colors in these images and you can train them in about two weeks to do just as well as a pathologist if you aggregate the score about four pitches they do just as well as a trained pathologist for breast cancer so this is the state of things now in this particular study the point was to say that pigeons are actually a better way of assessing the human performance compared to that of machine learning so even computers are not able to do what a pigeon is able to do so this is why we need to have better markers and what we're proposing here to do is to use the indulgence molecular composition of the tissue itself stuff that's already there effectively phenotype ng to enable you to have images that look like this were just by looking at the bio chemical composition of the tissue This is without staining this is all looking at the indictment composition you can barely clearly say which ones to melanoma and which one is not so for the rest of the talk I would describe different strategies that we're using right now to try to give you this type of information so that we can assess number one is there a disease there and number two how aggressive is the C's and how urgent is the intervention and what kind of intervention do we really need to have. So again the goal is to provide a health care protection with quantitative phenotypical or quantitative molecular information that can initiate treatment when it's necessary and avoid over treatment and over diagnosis when it's not really necessary. So first I'll go into detail about compromise spend the last few years working on this particular technique I'll touch on some of the other methods that have been proposed to be used in has the pathology. Imaging touch on our latest methods to try to extract again and dodginess molecular information from the tissues to assess the aggressiveness of a disease. So pumper microscopy is a non-linear method that interrogates the electronic excited states of molecules so these are the energy levels of the molecule and what you were used here is basically two frequencies and here I have a green and a red and the schematic here you have yellow. And red and you use these two photons to get you up to an excited energy level and then you control these two beams to actually interrogate how long the electrons are in that particular energy level there's a number of interactions that can happen and what you're monitoring is not new light that's being generated it's not fluorescents what you're monitoring here is changes in the amplitude of one of these two beams that gives you information about the dynamics of these electronic excited states what are the what does that BIOS well for melanoma we can actually get a very detailed molecular information from melon and that's a strong absorber so if you just look at the linear absorption you there's two types of militant you melanin a fair melon bilinear absorption of these two molecules it's pretty much the same thing there featureless so doing linear spectroscopy on this things is very difficult and you won't get a whole lot of information about these two biomolecules in a large study of about fifteen hundred patients they found that people with higher amounts of human element on average were more likely to have melanoma than those that have a higher percentage of pheromone and so the melanoma is has higher concentrations of human and in a particular study using leaner spectroscopy But what we really want to do is have very specific information at hire a solution and this is where microscopy gives us a lot more information here by looking at the dynamics of these electronic excited states we can look at the type of melon with really high contrast this is your melanin and fail melanin Riva have sensitivity to the oxidation state aggregate sites and metal content so we have a lot of specific bio chemical information of melons that there is no access to within or absorption. The way that this method works is by first introducing a pulse focused on to a sample so let's imagine that we have a very expensive in a fancy second light source somewhere downstream here that delivers two different colors the outputs are typically somewhere around one hundred fifty to three hundred femtosecond So these are very short process they come in and they're focused right on to the sample and what that does is it excites the electrons on the ground state the excited state as you see here then sometime later you come in with a different polls a different frequency and then it excites the electrons in this excited state some even higher excited state so these are the types of interaction that we're looking at and the key here is that this red photon is not going to be absorbed unless this blue photon precedes it so you need the electrons to be in this excited state for this molecule to interact with this red photon So that's the key principle behind pump microscopy you have this interaction with two different colors. So what we do in practice is we module that one of these two laser beams so basically we do this very very fast we turn it on and off on and off about Hertz. And so sometimes you have both courses coming in and sometimes you only have one of the two poles is coming in and so only when you have to to post this coming in together do you see this non-linear interaction so only at that point do you see this decrease in the amplitude of one of the two beams and when you lock lock into that particular frequency then you can detect these non-linear changes that's a result of this non-linear interaction. The last piece of the puzzle here is that we connect control the time of arrival of one post with respect to the other one and then that allows us to might the dynamics as you see here so we basically just a lay one post with the other one we do this multiple times and we're able to probe these dynamics at various time points again these are very fast time scales these are pico seconds. Analysis is done using principle component now so so basically we just take a large data set we throw it into P.C.A. and what we find is actually there's two or three different principle components that are able to describe the vast majority of the variance in that data set so in essence what we're going to be looking at is some type of ratio between the first and second principle component of the data and that's how we're able to construct these types of images so the color here corresponds to again some ratio of that first principle component and second principle component and there is some kind of correlation between the bio chemical composition but as I showed you earlier there's a bunch of different things that contribute to the signal and it's not one single thing it's not just human or pheromone and it gives you that signal it's all these different common factors combined. So here's one example. Where we have a patient with melanoma but in this particular tissue samples there's some regions that look fairly normal there's some regions and a half early stage melanoma and some areas that even have pre-cancerous States so not quite cancer so this is a very unique lesion because we have this very clear interface between a fully evolved on Noma region and a pick meant to normal region of the skin so what we do is we have an sample we show this to our pathologist they tell us this part is of interest this is what I would say this particular region and this then we look at the unstained slide and we image that with pumper microscopy so there's no standing in these particular samples and we just pick these ridges of interest that are pathologists has highlighted for us and then we image those areas. Right here just representative results so here you see a whole stuff happening but let me zoom in a little bit into first the early stage and the full the Evolve melanoma. So here what we see is changes in the bio chemical composition at the bottom of the Red Sea ridges these are. Basically this finger like structures that you see here the Malana sites under normal conditions should just be here at the base a layer of the epithelium But now because this is early stage melanoma you see this bio chemical composition that's not present in the melon and in the epithelium once you get the fully evolved melanoma you have this complete loss of biochemical and structural composition so everything's a lot more disorganized than what you have even in the early stage melanoma. I think one of the most striking images that I've come across at least in this label free imaging world is this particular case where we have this melanoma region. Interfacing this absolutely normal looking piece of skin and in this particular case even though they're both highly pigmented and the structure by eye doesn't actually look a whole lot different there is this very clear interface between where the melanoma is and this normal area of the skin so this is really really I think impressive. We've done this at this point with around a thousand different samples ranging from continuous melanoma ocular melanoma in general melanoma us so these are serious concerns from our initial Catania samples we looked at fifty five that was published in Science traditional medicine but by a combination of again the color of the bio chemical composition and the structure so the granularity the organization of the pigment in respect to the base a layer or the epithelium we are able to provide a sensitivity and specificity upwards of about eighty. Ninety eight percent which is extremely high and these are very encouraging results. Now we've taken the next step here which is can we now look at just an invasive melanoma not really say OK is this malignant or is this benign but have an invasive melanoma there's no doubt that this is an invasive melanoma and ask Is this a medicine addict melanoma or not and we do that by looking at the center on a biopsy or by looking at patients that have reoccurrences five years after their initial diagnosis and in this particular case these are fairly representative images where we see that there's actually a lot more biochemical or color differences heterogeneity than non-man a static sample so this is a negative sent in the biopsy patient compared to those that had met a static disease so here it seems that there's one type of phenotype that has persisted and the other very striking feature is the granularity of the pigments where things that are a lot more granular tend to be a lot more aggressive than those that have coarser types of melanin. All in all we were able to extract somewhere around thirty different features from these images and we do for future sequential I grow them to select the most important features from this which again are a combination of the bio chemical composition and the structure of the tissue and with that we're able to obtain a sensitivity and specificity of about eighty five percent So again this is how the encouraging particular considering that again the current specificity using the breast the thickness is twelve point five percent so that's the standard of care right now so this does much much better so we're very excited about these results. Now one of the problems here is that with pumper microscopy you're limited to strong absorbers which basically means hemoglobin an melanin Now we saw that with melon and you need this technique because linear spectroscopy doesn't provide this detail of bio chemical information with you can just use linear spectroscopy and you have a much cheaper system that can give you the same information. But so this is a very limiting factor that this is why we're focusing on melanoma only but if we want to look at other types of cancers we really need to have other strategies. One of them. Scattering which again uses exactly the same approach except for the lasers now instead of being found to seconds they're typically pico second or nano second and that's the have higher spectral resolution than high temporal resolution but here what you do with the two colors is that the color difference or the frequency difference between your two lasers actually go here only drives vibrational modes of these molecules you can get very specific bio chemical information there's two regions of interest one of those called the fingerprint region and the only ones the call the high wave number region they give you information from proteins Lippitt D.N.A. alas and college and it's very very rich in molecular information. With that you can get extremely nice images so these are Tom a graphic meaning you have three dimensional information because tsunami a technique you have a whole lot of molecular. Information and here they've color coded the nucleus college an act and again this is label free there's no labeling whatsoever of these molecules and so you can see that there's just a lot of molecular information that you have access to now actual quantitative results of looking at this particular technique and correlating that with the aggressiveness of certain diseases hasn't come out yet this is still a work in progress some people in this are actually looking at this right now but I think this is how the promising now similar to pump from across this technique of heroin and scattering requires very expensive laser systems the one that we're buying from my lab right now it cost about three hundred thousand dollars so this is a very sophisticated piece of equipment that you need for this and it's also point scanning mean that you're only going to be able to extract information at what point at a time so if you have large tissues to look at it's going to take days so you need something that's fast but you want that type of molecular information so other methods that have people have looked into to try to multiplex spatially basically they have to rely on linear interactions one of these linear interactions is scattering and it turns out that a lot of the scattering from the tissue comes from nano scale structures and those nano scale structures give you a lot of information that correlates with the aggressiveness of disease over the past two to three years people have shown that by looking at the nano scale structures by looking at scattering you can actually predict who are going to be progress or in prostate cancer basically those that had a low Gleason score that then later evolved to have very aggressive cancers or who had a biochemical recurrence after me so you've taken out the organ you shouldn't have any more disease but in fact you later have a spike in your P.S.A. meaning that you you have some type of medicine disease so you're able to predict those patients prior to the disease will be progressing with the nanostructures and it's also been shown to help protect them progresses for colorectal cancer so again scattering can provide a lot of rich information here these are still very early on. Studies but I think they show a lot of promise but one of the hurdles that they're going to come across is the fact that there is no actual molecular information so when you ask well what's going on the only answer they really have is well that's more disorganized that's pretty much all that they can say now they've been some early correlations to the Nano architecture and chrome it's in that I think it's very promising but I think when it comes time to translation into the clinic you really want to have more specific information as to what is actually happening what are the underlying phenotypes that demarcate an aggressive cancer versus a more I don't want cancer. So just to summarize here there's a number of techniques that I've introduced and they have it's tradeoffs positives and minus whether that is high cost the slow because their point scanning or the minuses being that lack molecular information one of the things that my lab is trying to develop right now is some way of bridging this information. Basically having all the benefits of these different techniques a way to have high molecular information specifically to a lot of different molecules in the body access to that nano scale structures the by a physical properties we want to be low cost and we want to be able to to multiply X. and space so that we can cover a large area is very very quickly and the way we thought of doing that was by going over to the alter Violet region of the spectrum. Now imaging in the U.V. is fairly challenging. When people have historically stayed steered away from it because it's harmful to the cells but if you're looking at fixed tissues there's actually not much of a problem when you do that and in fact if you go into the U.V. you have very detailed molecular information from trip to fan cytochrome C. collagen and then D.H. last influence and so forth so you do have that molecular sensitivity you also have higher spatial resolution so you can potentially get information at a higher spatial resolution you can also get the scattering information which gives you access to nano scale structures and so here are some recent studies that have shown the utility of U.V. imaging one was published a few years back now where they looked at two wave things at two sixty and eighty nanometers and by using some type of combination of these two wavelengths with a proper information you can actually quantify the nucleic acid mass and the protein at mass down to Phantogram So this is very exquisite molecular information that we know will already have a huge amount of utility for helping has the pathologist Now if you just look at a single way of doing a two sixty nanometers you can actually follow. All this information to look like in a chimney so you can even percent the data in a way that is already familiar to a pathologist which I think is really important for translation but again you have access to a huge amount of molecular information if you look at multiple wavelengths So the answer is to do some type of perspective basically looking at all these different way of things at the same time the way that hyperspectral imaging works is you basically shine your sample with a broadband source and then you use an Imaging Spectrometer to extract the molecular information or the spectral information. The problem with U.V. imaging historically has been that the cameras have poor quantum efficiency and the light sources have been very very dim but now over the past five to ten years this is actually changed quite a bit by looking at back eliminated CMOS cameras you can have a quantum efficiency of upwards of eighty five percent which is actually really good and now other news sources that used to actually give you very bright lights in deep into the the U.V. region of the spectrum the other challenges that persist though are this background fluorescence that typically is broader and has less specific molecular information within a certain band with and also chromatic aberration you have limited materials to make them the objectives and so chromatic aberration seem to be a huge problem in our objective that can give you a shift of more than about fifty microns which will basically prevent you from doing hyperspectral imaging so in order to avoid this what we've come up with leveraging in or from metric protection to try to overcome these two remaining challenges for looking in the U.V. with Enter from a tree what we're able to look at is instead of the intensity we're actually able to extract the complex field properties of the sample and then we can use this information to overcome any type of aberrations in the optical system and we're actually also immune to the fluorescent light so this biases everything we need in order to overcome the remaining challenges of imaging in the U.V.. So we can look at this complex field and we basically use interferometry or this complex information in order to extract the useful molecular information from the sample. So I'm going to go through a little math just so you understand how we're able to extract the scent from ation so just bear with me for a few minutes if we expand out this intensity term which is the amplitude squared of the complex fields we actually get this this equation that you see here that is composed of three different parts the first part contains the linear absorption so this gives us access to the total attenuation fish and this is what most of us are familiar with when you talk about absorption. The next part of this is called the linear phase modulation this gives us access to the roof for average refractive index and the thickness of the cell and it turns out that this is the term that enables us to have sensitivity down to about a man a meter or so for the samples. And the third term is the non linear phase modulations which gives us access to the changes in the refractive index as a function of wavelength This is known as dispersion and it's actually the key to opening up this whole thing you mentioned coefficient and being able to separate both absorption and scattering So how do we actually measure this well first let's just take the intensity for it's quite easy you just take the amplitude of this particular equation and you're able to extract this information immediately the more complicated part is this complex part and what you do is you unwrap the signal that again gives you two components a linear component and a non-linear component we estimate the linear component by fitting that to a line and then we subtract the residuals and that gives us access to the nonlinear components and again that gives us access to the changes in the refractive index as a function of wavelength. The key here is to use principles of salad where the changes in the refractive index are only correlated or related to the attenuation to the absorption coefficient not scattering So with that piece of information now we can actually separate out the two particular terms that are inside the total attenuation coefficients we get by looking at the Spurgeon or changes in the refractive index as a function of waving we're able to separate both absorption and scattering and uniquely sample these two properties now here just very quickly we did some proof of concept experiments looking at him a globe an absorption and trilobites a scattering from it which is this milky like substance that doesn't absorb light but it scatters it. And so we looked at this phantom First we looked at the absorption as expected if you just look at the total attenuation coefficient you're over estimating the amount of attenuation the amount of absorption in the system but if you look at the Spurgeon it is completely immune to scattering so you're able to reliably quantify the amount of absorbing from absorption from him of Logan in this particular Phantom and then you're able to get back to your scattering question so you're able to extract these two very important parameters in the penalty. Going back to this linear term here this is what gives us access to the typology of the sample and it turns out that we can measure those various with high sensitivity and we can make top of logical images of. Any type of sample of this particular case we have beats. Or sensitivity here is lower than one nanometer so we're able to protect changes in the apology of the cells in the order of a nanometer or less so here are some results are going into the U.V. region of the spectrum this is a red blood cell this is the amplitude and we focus around five fifty nanometers in this particular case if we look at the sample at that same plane at a different wavelength at three fifty nanometers the the cell now looks completely out of focus this is due to chromatic aberration but now because we have access to the complex field we're able to use the free space transfer function to then digitally correct for this chromatic aberration and so now we're able to extract something that looks nice sharp and in focus and we do this for every single wavelength to make sure that every single wavelength in our sample or in our measurement is nice and in focus and now we can extract both the changes in the refractive index from the cell and also the attenuation properties of the cell that is going all the way down to two fifty nanometers all the way up to about two fifty four fifty nanometers So you have a huge amount of bandwidth to extract the molecular composition from the cells the red line here is just a sanity check using solution phantom. We also did. Another test using a slightly more complex cell this is a neutral felt it has a more for nuclear structure meaning that it has three lobes for the nucleus it also has all these granules here that have various types of proteins. And so this is a much more complex structure than the red blood cells were essentially just bags filled with hemoglobin and in this particular case we see a lot of interesting structures from the nucleus we have this nice peak at two sixteen nanometers from nucleic acids if we look at the side of them we see this nice little shoulder around to eighty nanometers that corresponds to the proteins and we also see a smaller peak around four hundred nanometers and that's due to very likely model of Herat cities so again we have this huge amount of information just by looking at a broad band with. The changes in the refractive index here have less specific information but I think we just have to try a little bit harder to extract that information at the moment everything looks quite homogeneous was actually a bit surprising for us. We weren't expecting that. We can analyze the data in a couple of different ways we can use a power information to extract the mass of nucleic acids and proteins down to something Gramps again so here with a new map we see that this corresponds very nicely to the structure of the future feels while the proteins basically remains mostly in the side of Pozen we can also use other processing methods that use no a prior information this is called phase or analysis to basically map out the shape of the spectra that we have tamed to analyze the bio chemical composition of the cell again without any a prior information and we see two very nice clusters one is red here which are color coded and red and that corresponds to the nucleic acids here and then we have this larger area that corresponds to all the pro. Things so if you were to just look at this map you would say OK there's just proteins but when in fact there's a whole distribution that perhaps has very useful information for various different properties of different types of proteins perhaps. Now going first circle now and to return to his The pathology we've started doing some policy studies now looking at prostate cancer and to see here we have two sixty three ten three sixty four ten nanometers and these four images are from the same tissue but they look a little bit different signifying that indeed there is molecular information that's different at each one of these wavelengths being color coded this again and obtain something that looks like a chimney but this is fully quantitative it has very detailed information that could potentially correlate with disease and the aggressiveness of disease so this is very early stages particularly in the U.V. in the form metric. We've seen that there is a lot of promise from go ahead Roman scattering and nano scale sensing microscopy is a little bit more mature at this time and has a lot of promise but limitations here being that you can only look at melanoma because you're limited to melanin force in this particular technique. So hopefully I've been able to convince you that label free imaging has a place to improve and has a way of improving has the pathology and with that just like to acknowledge our funding sources from the US welcome fun and various foundations are collaborators from Duke University and also a Georgia Tech here and I'll be happy to take any questions. I think we do have to do clinical trials. And one of the first problems when we started doing pumper my cost before melanoma the one thing we always got is who cares just you're just going to cut it out so why even bother Right and so then we became more interested in looking at not just diagnosing the disease but staging it and asking deeper level questions that what pathologists are asking with the a Cheney or even him you know his the chemistry of slides. And so the translation aspect I think becomes very important to target things that current technologies are not able to provide. So even with digital pathology I think they're making really. That whole fields evolving very very quickly to the point that maybe they'll do better than a pigeon soon. I still think that looking at the phenotypical information as a marker of disease and particular aggressive disease is something that we can sell a stain perhaps So we do optical staining So rather than thinking of him in his chemistry stains with whatever market you're like we say well we're just going to do optical Seanie and the pathologists gets the same information they get an image that's color coded but now we can provide like with. Digital pathology a bunch of quantitative markers that say the D.N.A. mass is this the mass of the nucleus is this and so forth and all that information with the more clinical studies we can say this marker being this high correlates with this probability of progressing to stage four something almost Yeah. Yeah that's a very good question we did we naively thought that we were looking uniquely at. Your melanin family so if you if you notice if you look carefully the principal components actually look just almost identical to the signals that we get from your mail and pheromone and our Science Translational Medicine paper when we got this and then we actually looked at fossil data so if you look at fossil data the fossils squid NK in particular will have. Only human and no fail melanin That's because pheromone integrates over time you know and it's not and what we found were signals that look just like Pharaoh melanin on this Jurassic aged piece of fossils that we knew couldn't have any farewell and so we started scratching our heads and we said a what is going on here so that meter is trying all these different things oxidation state aggregates metal content all those also give us the same type of variations that we see between human and pheromone and so now we use a complete mathematical description of the signals that we see because you melanin with small. Looks positive here with small aggregate sizes will actually look negative just like there melon so there's no way of knowing uniquely what you're looking at we call them and members and we've done a whole lot of studies ranging from phaser analysis cake last serious principal component to this three dimensional graphical approach for principle complete analysis we've thrown everything at the try to figure out what's going on and the best we can do at this point is just to say that we have two and members that seem to be fairly consistent across all the different data sets that we've had. Yes So for looking at disease there's a cap of value basically how well do you agree with a pathologist or perhaps you had to have a panel of pathologist and actually. Pigeon study the main point of that was that you can potentially use the patients instead of the pathologist to do all these little studies and variations that there are a good surrogate for that but so the cap of value is one way to try to figure out some specificity it's actually a lot easier for the invasive studies because there you're not relying on the pathologist to tell you the answer. In that particular case you're just looking at is the symptom of the biopsy positive Yes OK So that's an met a static disease if no OK that that patient over the next five years the valid some type of recurrence if yes also met a static if not not met a static and those are very clear and markers that can then be used as correlations to our studies here so those have a lot clearer markers Yes in the diagnosis and saying this is going to be more of a Gleason score three or four or maybe a four and a three that's going to be much much harder to do so that's one of the challenges of this and other pathology take technologies. So. So there's very rare cases where we've seen that a unique type of bio chemical composition is an indicator of disease we've seen one case for. One of the more alarming cases of melanoma is actually genital mono most vulgar melanomas which are actually the second largest type of disease playing in that Anatomy it's quite churches but there is this one type of benign lesion that has a very unique. Pigmentation and that correlates uniquely with the raft mutations everything else that we've seen except for that one lesion it's not so much the composition an absolute sense but rather its distribution how heterogeneous is that pigment and then in correlation with its structure that's the more important marker and in fact people who are more pigment that it's it's almost even a better case because they have normal pigmentation then we can use this comparison so that is a proud mother problem with pumper microscopy that for more people with lighter skin tones there isn't a state of normal that we can compare to and a patient by patient basis now that's not going to be the case for these other technologies that I discussed where we're looking at other molecules but for melanoma people who are more pigment that is actually they have a better case for us to study them on us in an individual basis because we have a normal basis of comparison. Yes absolutely yes. So. So you're saying can we get it at a pre-cancerous state. OK So let me take the first part first which is why did. You let me try to interpret your question which is where does this fall under the current paradigm and where will it make a shift whether that is in a drastic shift or a subtle shift on the current way of the procedure clinical standard of care. And here we see this as an additional piece of the puzzle where again this would be like getting an immune has to chemistry stain where we confirm that there is the presence of this particular receptor in here week we will confirm that there is the presence of these particular phenotypes that have been correlated with aggressive diseases so that's where we see this fitting into is just one additional piece of information hopefully that strongly correlates with clinical outcomes that pathologists and clinicians can use so staging is not done just by the pathologist staging is done by yes the pathology but also a bunch of other clinical factors and so this is one of those pieces that's how we see it in terms of precancerous States I think the more exciting part to look at pre-cancer state is actually we're doing this in vivo not and pathology so that's a very different story and yes there is work that my lab is doing and that many other people are doing to try to figure out pre-cancerous States to even prevent the cancer from ever becoming cancer so you take it out before it becomes a problem but that is something that you're not going to solve by looking at his the pathology sleights So the strategy has to change quite significantly. Now this goes back to the problem of over diagnosing or. Over treating patients this is something that we're trying to avoid so. Yes identifying pre-cancer States is really really important but what we want to avoid is saying this patient is going to maybe possibly have an aggressive cancer so it's go ahead and give him chemotherapy or her chemotherapy that's exactly the type of strategy that we want to avoid and we want to say no no no hold on this patient does an intervention right now or perhaps different types of interventions. That in the end be better for the patient and for everybody involved. For precancer States I think in vivo imaging is the way to go and there are strategies to do that. That's a really great question so right now with our. Current system we're able to do. Each one of these little squares takes about six minutes. We can get that down to maybe a minute or two but this is point scanning for you and that's that's the problem. The nice thing about pumper microscopy and the one the ability to have a single marker that actually gives you information is that you can actually sparsely sample this and get fairly representative information so you don't need to have this very high sampling as long as you have somebody telling you look this is the area where you want to look you take just a somewhat limited area region of interest and then you're done when you're looking at scanning the whole entire tissue which is actually what I think needs to be done to be compared to be competitive with immunise chemistry or other technologies and this is not going to cut it and that's why a multiplex and specialty is so important. Yeah so I love the optical care fact the answer is I don't know but I'm really excited about it so this molecular and stations the optical care fact this is basically shining a laser with a particular poor station and things then move to that particular orientation and then they will diffuse this is basically the optical analog of M.R.I. The challenge there though is that I don't know what that would look like for fixed temples I would say that maybe fresh samples would be a better environment for that but for Fix paraffin and better samples I don't know if you actually get anything to move.