Gor Sahakyan, Yeva Yeghoyan | Bioimpedance Spectroscopy Non-Invasive Tissue Analysis

[music] Hello everyone, my name is Yava Hoyan and today my teammate Gorsa Gan and I are going to present our capstone project which is about bio impedance spectroscopy system for non-invasive tissue analysis. Bio impedance spectroscopy or BIS is a powerful bridge between electrical engineering and clinical diagnostics. Hence our goal was to create a system which will have a high precisioned uh output um which will help us to understand the physiological changes of the tissues. To give you an overview of today's presentation uh we will start to by understanding the core clinical problem and our target groups. Then we will introduce our hardware specific solution and the fundamental science of what is PIS and how it works. After that we will dive into the technical implementation of our design uh the block diagrams custom PCB and measurement results and we will conclude with the comparison of our device with an existing technologies analysis of risks uh as well as um our vision for the future iterations. [snorts] Um our primary um motivation was uh for this project uh lies in a significant gap uh in a current clinical monitoring specifically for cancer survivors. After treatment they uh face a high risk of the of getting the lympadema which is a chronic condition uh when fluid builds up in the limbs. Um in a traditional method it can be detecting uh when the fluid volume is reaching roughly roughly 300 ml uh and by that time uh the condition is already chronic and much more difficult to reverse and that will create a wait and see a state of anxiety uh for the survivors. Beyond that, our device serves uh for athletes for monitoring their body composition such as fat mass, fat-free mass, muscle mass uh and uh for tracking their overall hydration for their um body. Um our solution uh was designed uh to uh generate, inject and measure the electrical signals at a high precision. uh it should uh generate um electrical signal signals uh inject the current and also do a data logging. Uh as you can see it's uh it is a cohesive unit that integrates uh all of that um [snorts] >> components. Uh so um uh we have uh the data which we will get uh from the uh analog hardware that we have. Then this data will be sent to uh to the PC when our software will conduct the analysis and visualization. Unlike bulky equipment, our uh device will be portable, cost effective and will ma make the long-term monitoring for the health uh a reality for the people for on their own homes. So what exactly is bio impedance spectroscopy? At its core, it it is a measurement of the flow of an um of the body's opposition of the flow of an alternating current. Uh we are using this um frequency spectrum range uh from 1 kohz to 500 kohz and we are injecting a really low level current um to them into the human's body uh for about 400 micro ampers. um based on the a um ISO standards, we're not allowed to inject more than 1 milliamper which is al also uh very low for having the nerve stimulation or any tissue damage. Uh now let's understand how it works. The accuracy of our system relies on a four electrode or tetropolar configuration. We're using four four electrodes. Two for injecting the current and two for measuring the voltage drop. This is crucial because um it eliminates the impedance of the electrodes themselves. Otherwise, it can skew the results and uh the spec uh spectroscopy aspect of our system is that we are sweeping across the ranges of the frequencies. As I have mentioned, we're not using any frequency which is below 1 kilohertz because in that case the current uh penetrates only um in the extracellular spaces as you can see here. However, at high frequencies it can go uh to the both intra and extracellular compartments. >> Uh hello everyone, my name is Gore and I will talk about how our BIS system uh is implemented. uh as you have already mentioned we use the tetropolar uh electrode configuration and we can uh depicted the human body as a RC circuit with a parallel of R. So when we injecting the current through this circuit uh it happens the voltage and current uh phase uh changes and via the increasing of the frequencies this delay became higher and higher. So every point we can uh conduct and uh calculate the reactance and the resistance of this similar circuit in the different frequencies and uh get the result of the low frequencies for the extracellular and the higher frequencies for both extracellular and intracellular. Uh here uh for confirming uh our measurement method correctness we connect uh to our uh system capacitor and at the which is on the uh left side and you can see that uh we are increasing the frequency the delay between uh the delay between uh current and voltage became more visible. And at the right side I connect uh the system uh to my hand. And here you can see uh that it adds some noises because our hand is not the ideal capacitor circuit. And uh during the uh highest frequencies you can note that the uh phase uh between the voltage and the uh current changes as well. So this confirms the correctness of the measurement. Uh here is the block diagram for our system. Uh we use the controller for controlling the function generator. This function generator uh is the reference uh signal generator for our system. After that we need to amplify that signal and convert it to the current. Then we inject it through the body uh and use the analog front end with which is uh in this case the instrumentational uh amplifier uh this is the human uh body for the measurement devices and then we use the higher resolution highspeed ADCs for capturing the signal for the f far further analysis for the controller we use the STM32 uh blue Bill uh this control uh the function generator AD983 uh 33. So we also connect the push button it used to start the process. Uh we also has the LED. So when we but when we press the button LED turns on and indicate that the process has started and after the process is completed the LED turns off. uh as I already mentioned we use the uh 9833 function generator. This is the reference and this creates the signal from 1 kHz to 500 kHz with a sweep of 10 kHz step. Uh for the amplifier current pump and the analog front end we designed the custom PCB. Here you can see that we uh at the uh left corner the 3D model created in the Altium designer. Uh at the center uh we use the fabricated uh in empty PCB and on the right side you can see the fully assembled and functional uh PCB. This [snorts] is also act as the main hub of the hardware of the system because all the parts uh connected to it. Uh so uh this is why we uh create the custom PCB. Uh here is the schematic of this PCB. Here you can see the part of the function generator, the amplifier, the current pump, the instrumentation amplifier. Also here is the shunt resistor for the current measurement and respective the amplifier for that. Uh for the further iteration we also add some uh high resolution high uh speed ADCs. Uh you can also see that we use these uh jumpers to isolate each component uh from each other because this simplifies the troubleshooting step uh during our project. Uh for capturing the signal we use the analog discovery tree uh device. Uh this device has two differential oscilloscope channels which allow us to uh parallelly uh measure phase difference between the voltage and the current. For uh analysis we use the love view environment for capturing the signal uh and uh for the further analysis and also for data logging. Uh we also use this device for supplying bipolar power supply for our op pumps. uh plus - 5 volt. Uh for the analysis we use this call graph. Uh here you can see that the x-axis is the resistance, y-axis is the reactance. Uh we have a few terms R0 uh which represent the impedance in the lower frequencies. R infinity which represents the uh the impedance in higher frequencies. Here you can also see the term of phase angle and the characteristic frequency. Uh these uh terms speak about the healthiness of the cell membrane and they direct proportional to each other. So higher the phase angle uh higher the characteristic frequency healthier the cell membrane. So we use our uh measurement and uh create the call graph. Uh from the left side you can see uh graph created by our devices which measured uh my body [snorts] and we use the uh literature information. They use the similar setup uh with the similar frequencies with the similar sweep step uh [snorts] frequency and measure the unhealthy tissue which has the problem with the same membrane. Here you can clearly see the difference of the phase angle and also the characteristic frequency. [snorts] Uh it is much lower than the healthy tissue. Uh also you can see the difference of the R0 and R infinity parameters. They much lower than uh healthier ones. >> So as Gore has mentioned um we're getting the reactance and the resistance for the results. Uh after that the most important part part is phase angle. So higher um it is the healthier is the tissue. However it is important to note that uh where we should integrate more tools uh additional tools besides our device for getting the corrected phase angle uh to have fully precise result. Uh now we have the measured one uh which is very close to the actual phase angle but with the integration of additional tools we will get fully corrected one [snorts] and from that uh from all of that formulas uh we get the biological formulas that we need for example for body composition or for the hydration level. uh these formulas are validated from the company which is specialized in this field and have their own researches uh according to this um topic. So we uh took this formula to test our measurements. Uh and here we can see fat free mass, fat mass or even total body water. And the most important part of this is um the volume formula uh which is um uh which has the KB constant. As you can see this constant is crucial cuz uh it treats our body not as a uniform cylinder but as a combination of four cylinders and for the adults there is a general constant for that of 4.3. So we just plug those values and the personal reactance resistance or other uh body parameters and getting our our results. >> Uh here is the uh feature and the limitations of our system. I will mention a few of them. Uh we can say as a feature that it is noninvasive. So it is safe and painless. It is portable because we can use it in the home or point of care whatever we need. It is multi-frequencies uh compared to the single frequency devices. Uh so it has the higher accuracy. It is cost effective and it allow us to make the realtime uh monitoring. uh as a limitation uh it requires care careful calibration protocols uh and it needs to complex data interpretation also uh bad uh connected electrode can affect by uh un unexpected results. So that's the features and the limitations. Um now here we can see the compare and contrast table for our device and two other devices which have some similarities with our device. The first one is the smart scale. The smart scale is based on the single frequency analysis which is mo mostly statistics rather than personal um output for the person. Uh the last one is uh imaging machines which are very good if you want the detailed anatomical details of your body. However, our device is something in between. It is cost effective and portable and it can give you also the details of your body, not too detailed, but the hydration level or body composition if you want to track it uh on a daily basis. Um here are some risks that we have for our device. The first and foremost one is that um our device is in its initial stages of the development and we do not have the error handling yet. So if something breaks inside of the system, the user will not get the alert of this and this will lead to a need for the recalibration from the professional repeatedly. Uh also there is a risk that even though we can uh detect the condition from the early stages, the doctors may not act until it is visible for them. uh and also that the customers can uh confuse our system with the treatment and miss their um visit to professionals. Uh but it it is important to understand that it is not a treatment or something like that. It is just monitoring uh if you want to do it regularly and it is important to have your treatment uh outside of this device usage. uh about the future development and the iteration uh we consider to change analog discovery tree with the high resolution highspeed ADCs. This will allow us to reduce the cost about 80%. Uh we also consider to add uh some lithium ion batteries and use the mobile bio application for for the data interpretation. This will allow us to uh make the device more independent from the PC uh or the laptop and we as use uh as we use the 3D printing for the prototyping the case we plan to change in more durable one. So in addition uh during the poster presentation uh some people from the oral uh approach us and uh make an offer for collaborate collaboration uh they made the device uh of high voltage uh way uh to inject the uh cell membranes. So their device uh don't have capability to measure the impedance and our device has. So they plan to collaborate and integrate our device to their systems. >> Um to conclude, we have successfully built a device which is non-invasive, portable and cost effective for monitoring every health metric that you need in your everyday life. >> Thank you. [applause] Great work. >> Thank you. >> And uh can you go with the slide back where you have the the problems list that you have mentioned? >> Right. >> Risk. >> Yeah, this one. >> Have you done some risk mitigation study of like for the listed items? Any thoughts on how you can mitigate in a next iteration for the device? Let's say >> uh how we add uh >> uh what we will do at the first stages. >> Well, no, you you have the device right there, right? And you have the list of your technical risks. All right. >> So, have you done any mitigation study of how can you in the next iteration let's say fix or partially fix or resolve any of these risks? So we can partially fix the risk of for example uh the error handling part because we can integrate the hardware and the software part uh to make uh device every time test uh self and after it gets an error it indicate the user there is something wrong with the device so you can't use or it need to uh it need calibration uh before using. So that part we can eliminate uh in the future iterations >> also maybe when it will be medically validated so doctors will trust the device and will take some measures to um do something uh for preventing the condition to the worsening situation >> and also you mentioned one of the existing devices is using statistical data. Not one of Oh yeah. Uh the smart scales uh cuz they are based on the single frequency which cannot um have the precise information it is working on the one frequency range. uh and uh they are using the statistics. For example, at this gender and age um the normal for the men is for example 45 uh kilograms and for that um weight you should have uh this kind of fat-free mass but our device uh analyzes um a lot of range of frequencies and it is more precise and more personal for that person. So what I'm think what I'm saying is that there's a clear evidence that the statistics brings the value to this right. So have you thought about adding the statistics on top of what you have done? Is it possible to do or have you thought about it? >> Uh for this device uh there is no need for the statistics to be implemented cuz it it has the ability to measure the intracellular parts which is the most important one. the single frequency ones do not have that ability that is why they are using the statistics. >> So that doesn't make any sense to add on top of what you have statistics. uh after the getting the outpost when we use the result part. So five months uh there we use the statistics of the uh >> yeah of men and women uh for the same company uh from their researches to compare which values are normal for the healthy people people and for the unhealthy ones. So we used statistics there. Uh well given that the device is primarily oriented to pet customers we have some clinical condition. >> Mhm. >> In the introduction we mentioned cancer survivors or post cost post treatment dehydration realation issues that is tell to solve. I think uh you should think about uh we are uh making a variable product >> for the patient to carry on to himself for herself. uh but that your device parameter doesn't help. >> Mhm. you mentioned in the papers I did books uh between 15 and 30 Celsius whereas the patient can have up to 40 41 and we should think about it because if we are looking at the uh immediate changes in hydration levels >> which the patient has to pay immediate attention then uh this has to be taken in consideration >> of course of course >> as uh development advice. >> Mhm. Thank you very much. >> Small question. >> Mhm. >> Uh probably you mentioned but uh regarding the analog circuit as probably most of it is about the PCB as the device you said is quite should be very accurate. So the measurement should be precise you will have area for errors. Have you done any measurement or do you have any considerations of having some of reducing the interference of the other components on the under circuit or adding some shielding or simul because the sensitivity is very important. uh because this is primary primary the P of the concept uh we take the into the consideration uh and the future development we will add some circuitry to uh add the accuracy of the overall system and also uh make the device more precise uh when it do some measurements >> and uh it will to >> to the experts that maybe the offices say interested to have this monitoring but when we say this monitoring is mean that during the time you have measures and this measures should be somewhere to understand the dynamics etc. Is there any interface let's talk about what docker would like to see >> and what is the timing in the time scale how you provide the values >> chance probably >> uh as we already mentioned with some interface we use the lab view environment uh >> I know love but love is just for capturing and providing the >> but this also can uh we also love you for logging this data so It can be used in the further investigation of the problem for the monitoring or am I understand your question? >> I think so. But doctor will not accept it that lot. >> I understand that if you're collecting it, capturing it and one day you will provide it the >> of course we plan to change the analog discovery with the high speed high resolution ADCs. So uh for the further uh implementation we will use another software and uh for capturing and data loading. >> Okay. >> Did you tested it on several people or >> we test it on me >> only on? >> Yeah, we're not permitted it. I think one one thing is that uh maybe you need to have a supervisor like someone from uh medical field to understand this feed but also the the skin thickness the color of the skin everything will impact the accuracy of the result that you have. So that's important you test on several people and then maybe uh you need to make changes on the design amplifier and things like that. So uh if you do more test again you will get more accurate uh of course design and result. >> And if I can just add to that even calibrate against other data >> approaches in other words you can get a sense of how what your variance is for a given person compared to alternative sensitives. >> Mhm. >> Well also directions to maybe ground your work a little more in context. Even the moisture of the skin can cause the >> right but instead of qualitative judgments like that you can actually get >> so how the variance and and uh you know what's the standard deviation or some basic statistics on >> I did 100 measurements morning noon night dehydrated hydrated I ate I didn't eat I don't know whatever all the factors that could be Thank you. >> Thank you. [applause]