 Thank you very much for coming today to the biohacking village speakers and be able to start this off pretty soon but definitely first and foremost we want to thank all of you especially coming this early on a Saturday morning in Vegas so really appreciative of that and thank God for Tylenol. Definitely first and foremost as well we really want to thank our sponsors for this year because as we've been growing as a village we've had wonderful people be able to come in and donate with us just to start off with a few of our bronze donation groups Medisau, Digisert, Electrolabs, ICU Medical and Intel, up to our silver the Megcrypt and Siemens, our gold group Metatronic, BD and Thermo Fisher Scientific and our platinum Donators Abbot. So definitely thank you all for helping us expand out and get even bigger from just speakers to now being able to do wet lab which is over in Melrose one and our super fancy medical device testing over in Melrose three. One thing as well if you would also like to make donations to help the village grow even more we have wonderful donations set up in the back with my good friend BJ back there. With those donations comes wonderful swag so you should definitely check it out. We do ask that you do not go back during the talks we will actually not be taking donations to be respectful of the speakers but we will have breaks in between talks so if you are interested to go check it out then. Just for reference as well with the wet lab we will also be having a chip implants and that should be at one o'clock if you are interested or if anyone you know may be interested please swing by over to Melrose one there will also be an updated schedule and then finally for here just to let everyone know because there are a lot of groups wanting to come in to see all the talks we do rotate out the room so after this talk we will ask for people to go back out we will have a line that you can jump right back in but just to be fair to everyone else we try to make sure to allow everyone to be able to sit and go through all the talks and also for questions we will hold off until the end we will have time allocated but if the speaker does run over the speakers are more than happy to take talks outside and without further ado just the USBC and and now we'll have our first speaker of the day Jin Ritto wants to push forward a health technology commons previous experiences include bringing consumer electronic biosensor products to market from the emotive BCI to the basis watch and kiddo biosensor watch and being published for her work in cognitive neuroscience and nature let's welcome our next speaker Jin hi everybody um thanks biohacking village for inviting me to speak today it's an awesome opportunity deafcon is pretty fun um today i'm going to talk about um my efforts to uh make an open source biomedical imaging device um and why i think that's important for moving healthcare forward um biomedical imaging happens to be in this interesting intersection where you can possibly do it without cutting anybody open uh you can do it not invasively um so in a way you can actually just innovate the technology um to look inside the body so innovating technology to look inside the body is actually something that anybody like you might as well have a go at because uh you're not making a diagnosis so um that makes it uh not a medical device um and you can collect the data the data happens to be very useful um and for for moving forward healthcare um so i'm going to talk about um how to make medical imaging more accessible to more people just so that you're well aware um only so two thirds of the world has absolutely no access to biomedical imaging devices right now um which is pretty shocking and pretty terrible because we use them all the time to diagnose various healthcare problems so even in the united states of first world country you'll find that we have um you don't hack with a biomedical device very often unless you have a really big budget um in fact you conservatively get scans and they often cost you extra money out of pocket um when you go and get them so there's some scans like an MRI scan which is definitely a fantastic high resolution scan but you probably won't get it most of the time unless they already know you really need it there's something that's very suspicious and they just want to confirm it um and i think it'd be super valuable if we could just be doing scans safely more frequently for preventative measures and um we need to figure out how to get there so um my twitter's mines i bio um just to give you an overview of what's going on with biomedical imaging right now um we've got CAT scans which a lot of you have heard of they send x-rays through at different angles um and uh unfortunately they're ionizing radiation which is bad for you um they cost a few million dollars um so ionizing radiation is um causes cancer over time over exposure in adults they need a lot of maintenance definitely um a cool item uh very large you probably won't play with one at home both because it's dangerous and huge and expensive an MRI beautiful physics amazing now it's about three million dollars for a standard MRI scanner right now oddly enough they haven't come down in price since they were invented which i think is a very sort of interesting uh point to note um most electronics do come down in price over over decades in consumer electronics particularly so why hasn't an MRI come down in price well to use an MRI you need a giant helium quenching chamber for one you need a giant electrode magnet um it's a really pretty big investment again it's got great spatial resolution but not very good time resolution so each of these modalities has different pros and cons about them they're good at different kinds of imaging ultrasound um ultrasound a medical grade ultrasound is about 115 thousand dollars on average um it has problems due to scattering and any EEG uh 40k for a medical grade EEG you can currently uh it doesn't do any spatial reconstruction at all so it's not a medical imaging device but it does have an interest it does get very interesting sense of information so i think it's worth mentioning there is an open source effort in EEG called open BCI um so the open source movement is starting to enter this realm but how far can we go so just as person like computers started out taking big rooms um we shrunk them down to the personal computer size and then we shrunk them down again to the cell phone size and we've had the personal computer revolution everybody feels empowered we have conferences like def con and i don't think we would have this type of get together with so many people unless computers were small and accessible to everybody so i would love to see the same thing happen for the MRI scanner look it's taking up a giant big room here and look Star Trek tricorder that's what we would like to get to a tiny biomedical imaging device so today um i just talked about why it would be useful to make medical imaging more accessible um i'm going to talk about what the biomedical imaging project i've been working on is how it works what it does and what you can do with it and some applications and then i'm going to finish up with talking about how i think it ties into a pathway for a new patient centric model for healthcare innovation i think this is very important to sort of bring up um at the end um because uh lots of people seem to be unhappy with the US healthcare system as it as it is and don't really feel like it's working for them as patients as and consumers and so there's some big questions about how do we get alignment and how do we make a system that works for us so spectra uh open biomedical imaging so it's all open source and um if you would like to see the code um all the repositories um you can go to open eit.github.io um it's a device that's like EE firmware and software um but this entry page has some tutorials and installation instructions to help people get off the ground so you might be asking erica great what is this kind of biomedical imaging that you're talking about i'm talking about um a technique called electrical impedance tomography which is used um historically only in a few labs actually um it's non-invasive sends a very tiny amount of current through the body it's non-ionizing which makes it safer than an x-ray or a CAT scan it's compact and inexpensive um because uh with all of our consumer electronics we've managed to make our current sources and precise current control is something that we can absolutely do whereas if you say look at ultrasound and an ultrasound transducer with impedance matching required to make an ultrasound transducer they're so quite expensive so because we have innovated so much in our control of current we're very good at this and it has a good time resolution and it has better source localization than EEG um nobody knows about it that's not so good and the spatial resolution is limited by the number of electrodes as well as the difficulty of the inverse problem which i will get to in a little bit on the right are some typical images that you would get from this modality of imaging um you can see um that's a phantom on the top right because um which is a biomedical imaging test setup so you can see it's reconstructing the position of the cups in the tank um and then we've got a baby here and electrode wrapped around the chest and it's able to do thorax or lung imaging so you might be asking why specifically use current um which is a good question because there's lots of different types of energy that you could shoot through the body and each of them actually has different pros and cons and potentials with regards to biomedical imaging and i wouldn't want to um limit conversations about any of them but i'm going to focus specifically on current today so um if you think about the body it's actually a grid of it's a circuit and the circuit is made up of cells and the best way to understand eit is to look at what an individual cell is it's an electrical circuit made up of resistors and capacitors the diagram here shows um the um intracellular and extracellular fluid modelled with a resistance they'll have different resistances and the cell membrane is capacitive now the characteristics of a cell populations plasma membranes and volumes and these conductivities influence the impedance spectrum um and the impedance spectrum is obtained by uh running a frequency sweep uh or applying various alternating currents between 100 hertz and 80 kilohertz um to get the dielectric properties of tissue when different frequencies are injected through a cellular medium they take different routes to find the path of least resistance thus different frequencies will measure different materials such as extracellular fluid versus uh versus cell membranes and their contents leading us to different spectrums from different materials this is actually pretty good if you want to distinguish between different materials in the body the change in impedance magnitude between the voltage receivers and the current emitters is the measurement that you use um in eit to recreate the image in the above picture you can see the low frequencies around about a kilohertz go around the cell membranes and at 50 kilohertz it goes straight through and that's very important if you want to measure one and then the other um this information gives a bio impedance spectrum um which we can determine the subtle changes in the dielectric material um this is a pretty cool technique on its own but it's made even more powerful when you combine it with tomographic or spatial reconstruction so how do you get an image so eit systems apply these small alternating currents at a single frequency or multiple frequencies to obtain dielectric spectrums at localized points in space small alternating currents ranging from a few nano amps and up are applied with the resulting potentials being recorded from other electrodes the process is then repeated for different electrodes configurations so that you can get every electrode configuration the basic of the imaging reconstruction was laid down by Johan Radon by making a measurement between all these different sets of electrodes we can infer what's happening between them so each of these impedance measures is a called a projection and then we perform an inverse radon transform to reconstruct the image the more measurements you can make the better the image reconstruction is going to be so here we have three projections and you can see a very grainy or a low resolution image but the more projections you can add to it the higher resolution it's going to get the device that i've built has 32 electrodes which is enough to clearly see that you can recreate an image and localize points in space if you compare that to a CAT scan a CAT scan shines through about 256 projections and then it rotates those as well so it would make about a thousand and twenty four projections to recreate an image which is many more that's also a lot of x-ray exposure one limitation of the back projection algorithm is that it assumes current moves in straight lines which it does not um so they've come up with some better models using using finite element analysis called the Gauss Newton and Graskin consensus algorithm to reconstruct an image um so in labs EID has been limited to these big carts of equipment excuse me um so um very little work has been done to actually shrink it down but that's what i've done i've made a much smaller version which is two inches square um and this is an example of what the reconstruction looks like um there's there's a cup and you can see as i move anti clockwise you can see exactly where it's located in the image this has a lot of applications and cheap diagnostic sensors um this is what the physical hardware looks like um so here there's two cups um and you can see them reconstructed in space this is what the pcb looks like you can do it in a tank or you can wrap it around the body part open EIT comes with free software to do image reconstructions in real time you can implement the image reconstruction three different algorithms bio and patent spectroscopy or the frequency sweep method or just do a time series impedance measurement which is quite interesting on its own um it's compliant with IEC standards which makes it safe for use on humans which is important um and it has bluetooth i should have a video here um one second so this is back when it was eight electrodes which is lesser resolution and you can see i put a shot glass in there and it reconstructs the position in the tank there's also gestural control um which i can also show you in the workshop which is a great application of it as um then you can actually do spatial localization of the muscles in your arm which is something that you can't do with EMG you can get breathing and heart rate which is pretty awesome because with say measurements you're used to which is just a voltage measurement you would not get the lung volume it so you won't get the lung expansion and contraction and when it's measuring heart rate it's not measuring the typical q r s t signal it's actually measuring something different it's measuring the impedance signal which is related to the um the expansion and contraction of blood flow in your heart so it can get the exact volume of air that's entering and exiting each lung if you wrap it around the body part say your your your chest you can get a blurry image of your lungs so you might say it's not as good as an MRI yes it's not as good as an MRI but it has better time resolution and it's accessible so this is what one of the earliest MRI scans look like and this is what they look like today this is actually my head um so um yeah with a three tesla scanner i used to be an MRI technician at a lab and that's what an early EIT scan looks like and the question is how far can we push it and where can we take it to how how good can we make it um i think this is a very exciting application of EIT just to explain what this is um i've done this is multi frequency EIT so you've you run through different frequencies and you do an image reconstruction at each frequency that gives you the different dielectric spectrums at each pixel in space um you can apply it in like a basic classifier i did a Pearson cross correlation here which is very basic and what's cool about that is here i put in apple a sweet potato or both in a tank and i'm able to detect where the apple is and i'm able to differentiate the apple from a sweet potato how do you apply this to medical imaging replace the word sweet potato with cancerous and non-cancerous tissue and they're already doing that actually and i'm going to get to some applications that people that are currently in research for this technique so um spectra has um i mean mcu it does these three different techniques um it has four multiplexers that give you the 32 electrodes um as a tank it's bluetooth enabled um it does three different algorithms it's all open source um and it's safe um i'd like to say thanks to um different open source collaborate collaborators particularly mariam ben one um who's responsible for some of the algorithms clement um who helped with the flex PCBs john nullty sebastian andrew and poll so i'd love it if you wanted to spread the word um i've got a website you sign up you get science the updates um even buy a kit which is basically the open source project assembled and uh put together ready to install so developing world applications at the beginning i mentioned that two thirds of the world has no access to medical imaging so what if we could just pick something easy to start with tuberculosis is a massive problem in africa um and it's i've got a cough right now but it's like a mucous lung uh mucous modules on the lungs so what if we could help by giving them some tool to diagnose this that would be awesome um one of the problems here that i'll just like to sort of point out with the system and how it stands is that there's not a lot of funding to help people in africa because there's not a lot of money to be made by helping them so this isn't a major motivator for people um who have money generally unless it could become one so i was aware of this and i thought maybe a better approach is just to make this technique available so people can help themselves um so here's the pulmonary edema example we're looking at the lungs pulmonary edema is water on the lungs so up there on the top right you can see the lung volume in each lung and you can see that in real time so that makes it very easy to see when somebody has water on the lungs so alongside this lung volume perfusion you could do bladder fullness you don't just have to put it around your chest um your bladder is a big conductive blob of of urine what about your stomach a fetal heart rate monitor from the outside i already mentioned the tuberculosis but um another good one could be ischemic or hemorrhagic stroke in the field so if you need a field sensor and another angle is non-invasive electrophysiology um as a replacement for EEG or an upgraded version of EEG so i'm going to talk about what labs are doing with this so um since spectra makes cross sections of limbs is perfect for experimenting with general control and new types of human computer interaction particularly people that struggle with prosthetic limb control um and they're typically just using a voltage measurement of the muscle movement and they can't accurately localize which muscle is coming from where EIT yields more information than what is traditionally used as it attains both spatial information and this dielectric information at different frequencies telling you what material is where there's um our researchers at Carnegie Mellon University looking into this and they made um a gestural controller with it um MRIs are the best method to diagnose something like breast cancer but they're rarely used you might ask why don't we use MRIs all the time because it's too expensive to even just do scans with them unless you really need to so they're not used for these kind of standardized scans but we could use EIT and there's actually a lot of research on EIT um looking at applying it to breast cancer here's another one um this is a company in the united kingdom replacing the pap smear where they take a sample of tissue they send it away to a lab and then it comes back a few days later with a cancerous or non-cancerous diagnosis what if we could do that on the spot without removing any tissue that would be much better and a company is doing that and this is going through clinical trials they're using bio-impedance spectroscopy um just because it's awesome and fun um brain imaging so EEG has good time resolution and really bad spatial resolution MRI has great spatial resolution and terrible time resolution how can we ever describe the complexity of the brain unless we have both so EIT has good time resolution and okay spatial resolution so perhaps it could give more data than EEG um there's actually some um advances in labs five minutes okay um in doing that and they've put ecog arrays on rats and are able to spatially localize action potentials here's another lab doing it as well so this is a fun thing happy to talk about that later what's the best that you can do with it 200 microns and two two milliseconds so that's like spatial and time resolution pretty good um I do think biomedical imaging is the path to better BCI um there's a company in San Francisco um called open water that's also making another medical imaging device um but making a better biomedical imaging helps all of healthcare along the way lab on a chip cell analysis you can do it at any scale just change the spacing between the electrodes uh smart bandages for burns or for detecting bed ulcers before they happen pressure sensors determine the pressure distribution of a surface like how you're holding a baseball bat or maybe you're a diabetic with um foot ulcers a common problem and you want to get a pressure map of your feet again uh infants can't go inside a CAT scan so do this um so now I'm going to shift to new models of healthcare innovation can we get on this path by making an imaging tool more accessible how do we even get to a patient centered healthcare um currently instead of seeing poorly aligned you've got your health insurers and they might be incentivized to help you or they might be incentivized to save money it's hard to say and not everybody completely trusts their incentives so an example of how things could go wrong there's a large medical device company buys another small startup and hides their technology because it threatens their main business that's happened um if we could share the underlying technologies in health which is so so important we could learn for uh we could learn faster and move this technology faster but how do we structure that with the current medical system um the healing power of your own medical records this guy um went to a lot of effort to get access to his own medical data and from that he was able to diagnose his own brain tumor which was previously undiagnosable if you did have access to your own data the question is would you look at it more would you experiment with it because you know if we get a whole lot of smart people experimenting with it something's going to come out of it if you saw something odd would you find out more like that's preventative healthcare right there maybe you'd try some new machine learning algorithms on it like make a Kaggle database of lungs or something um speaking of which um google came out with this google health using CAT scan data they are on lungs um their AI outperforms radiologists which is kind of awesome in a way um except that is hard to get access to a CAT scan so what if you could apply machine learning to electrical impedance tomography I don't see why you can't in fact there's some really great new papers in deep learning that um show great results at improving spatial resolution using machine learning so I've been talking about creating new data which is um kind of more of a risk and you might see this as more of a risk because it's like all this new data out here um however I'm also not talking about aggregating it and putting it on the cloud necessarily but I would like people to aggregate it to do machine learning to be able to make better diagnostics so what about patient privacy currently the data in your MRI scan is owned by hospitals and health insurers and it's actually really hard for you to get hold of it you can you have to go through the special medical records offers and you got to go through a process if you had a personalized imaging system it's really clearly your data to do whatever you want with your choice um so I would argue that decentralization of the base imaging tool will lead to individuals owning their own data and doing more interesting things with it so I would love to make medical imaging more accessible is there a way that this can fit into existing business models um can we grow from each other can we aggregate the information can we build an imaging tool together that surpasses what's in hospitals wouldn't that be a wonderful world of preventative scanning um anyway I'd love your support suggestions um and um the way that I I see it just to keep it in perspective and to see how much things have improved um I'd like to remind you that a hairdresser or a barber and a surgeon used to be the same profession and they used sores and knives so if something got infected they just lopped it off we've come a long way and it's awesome we now do keyhole surgery pretty good but the way that I see is going is actually bioelectronic medicine where we non-invasively like uh will kill like ablates tumors and we'll also get sensor information we'll read out and this is going to help people get better much faster and we're just at the cusp or the dawn of this technology pretty much finished so I've got a workshop today um in about 30 minutes in fact and I'm going to do some demonstrations with spectra um I'm going to show you the basic underlying technique do some imagery constructions do a bio-impedance spectroscopy demo talk about uh technical next steps and different ways that you could make the image better and we can have a question and discussion session and I might also do like an electronic architecture review as well for anybody who's interested so um that's my talk on making medical imaging more accessible and the spectra open open eit device um check it out um here are my details and I'm happy to answer any questions that anybody might have about it um yes uh you mentioned it was low cost what's the cost currently in terms of like setting up for yourself or getting the kid and how do you see it dropping sort of longer term as you take advantage of economies of scale um sure um well you can make one yourself but if you want to on the open like just go to the github and you can see all the detail there um if you don't want to do that you can pay 350 bucks and I will send you one um which is already assembled and ready to go with firmware flashed and works um and economies of scale this is like the the same consumer electronics question um basically they get cheaper the more um the higher quantities that you can buy uh you you mentioned it was low voltage is there is is it can it be powered off of a knife after a mobile device like a usb powered or would it need external power i would love to see like an app or something that would work with this and like a usb uh device um yeah um it's that's uh it can take a five volt input and it can be charged by a usb if that answers one question but all the main chips on it run at three volts they're in their low power precision microcontrollers from analog devices yep thank you