1 00:00:00,000 --> 00:00:15,640 *Music* 2 00:00:18,160 --> 00:00:24,940 Herald Angel: And now we come to the talk entitled low-cost non-invasive biomedical 3 00:00:24,940 --> 00:00:32,000 imaging. Current medical imaging has problems: it is expensive, it is large, 4 00:00:32,000 --> 00:00:39,940 rarely preventively used and maybe you've heard of the story of a fMRI - this is the 5 00:00:39,940 --> 00:00:45,960 magnet resonance tomography - they put in a dead Salmon and they can get a signal 6 00:00:45,960 --> 00:00:51,729 from brain activity from it. There's also lots of problems in the software as well. 7 00:00:51,729 --> 00:00:59,450 A little story, maybe you look it up. And how this whole mess can be solved with the 8 00:00:59,450 --> 00:01:06,769 technique called Open Electrical Impedance Tomography - this will tell us Jean 9 00:01:06,769 --> 00:01:10,509 Rintoul. Give a big round of applause for Jean. 10 00:01:10,509 --> 00:01:18,541 *applause* Jean Rintoul: Thank you. 11 00:01:18,541 --> 00:01:20,939 Hello everyone. Today I 12 00:01:20,939 --> 00:01:28,159 will be talking about an open source route for biomedical imaging using a technique 13 00:01:28,159 --> 00:01:34,090 that's in R&D called Electrical Impedance Tomography. Not many people have heard of 14 00:01:34,090 --> 00:01:42,609 it, which is why it seems like it's important to mention. First of all, I'll 15 00:01:42,609 --> 00:01:48,789 just give you the vision of what it would be like if everybody had access to cheap 16 00:01:48,789 --> 00:01:56,210 biomedical imaging. Right now you only get imaged when something's gone wrong. And, 17 00:01:56,210 --> 00:02:02,189 moreover, you only actually get to use these tools when something has gone wrong 18 00:02:02,189 --> 00:02:08,780 in a first world country when you're lucky enough to be close to a hospital and have 19 00:02:08,780 --> 00:02:15,030 access to these technologies. That's a very limited number of people. What's even 20 00:02:15,030 --> 00:02:20,720 worse about it: is it's hard to hack! So, if you wanted to improve this technology 21 00:02:20,720 --> 00:02:28,260 yourself - medical physics is an amazing field - but it would be very hard to do so 22 00:02:28,260 --> 00:02:34,510 because you don't have a three million dollar MRI scanner sitting in your garage. 23 00:02:34,510 --> 00:02:40,360 Maybe you do, that's good for you, just not many of us do. If we did have cheap 24 00:02:40,360 --> 00:02:46,190 biomedical imaging we could do things like do preventive scans so you would wake up 25 00:02:46,190 --> 00:02:51,520 in the morning you'd like, take a shower, the device would be quietly imaging your 26 00:02:51,520 --> 00:02:55,750 body, would warn you if the slightest little thing when went wrong. You'd do 27 00:02:55,750 --> 00:03:02,290 machine learning over it, it'd be wonderful wonderful for health care. So, 28 00:03:02,290 --> 00:03:07,750 that's the vision of what biomedical imaging could be. And the other point is 29 00:03:07,750 --> 00:03:13,310 sometimes we move forward faster when we share the information. I worked in defense 30 00:03:13,310 --> 00:03:16,410 for a brief period and people didn't really share information between each 31 00:03:16,410 --> 00:03:21,440 other, and I think that inhibited science from moving forward. So, sharing is 32 00:03:21,440 --> 00:03:24,760 caring. So today I'm going to go through a few 33 00:03:24,760 --> 00:03:28,020 different things. I'm going to go through the current biomedical imaging 34 00:03:28,020 --> 00:03:31,020 technologies. I'll give you an introduction to Electrical Impedance 35 00:03:31,020 --> 00:03:35,590 Tomography. I'll go through the open source Electrical Impedance Tomography 36 00:03:35,590 --> 00:03:40,660 Project. Then I'll go through some applications that we could apply it to. 37 00:03:40,660 --> 00:03:45,040 And then I'll suggest a few different next steps that we can go into because by no 38 00:03:45,040 --> 00:03:52,980 means is it finished. Right now we have four different main existing imaging 39 00:03:52,980 --> 00:04:00,910 modalities. Your MRI scanner, which is a wonderful tool, it's huge, very expensive. 40 00:04:00,910 --> 00:04:06,320 The most commonly used imaging is actually CAT scanner which sends our x-rays through 41 00:04:06,320 --> 00:04:11,070 your body which is ionizing radiation, which is bad for you because it causes 42 00:04:11,070 --> 00:04:17,220 cancer in the long run if you get too many of those scans and it's actually the first 43 00:04:17,220 --> 00:04:21,460 first scan that you'll get when you go into the emergency room. It's the most 44 00:04:21,460 --> 00:04:24,650 commonly used. And as we all know we've got those grainy images that come from the 45 00:04:24,650 --> 00:04:32,400 ultrasound of fetuses, wonderful tool except for the scattering due to the sound 46 00:04:32,400 --> 00:04:36,299 gets scattered when you have different density materials next to each other. And 47 00:04:36,299 --> 00:04:45,160 not exactly an imaging modality but a very important diagnostic technique is EEG. 48 00:04:45,160 --> 00:04:50,980 So you might ask, how do we classify these right now? we have 3 main types of 49 00:04:50,980 --> 00:04:58,850 resolution. Spatial, contrast, and time. Spatial resolution is, basically, what 50 00:04:58,850 --> 00:05:04,010 space you can determine 2 different objects from each other. Contrast 51 00:05:04,010 --> 00:05:09,000 resolution is soft tissue or subtle differences in tissues. And time 52 00:05:09,000 --> 00:05:13,540 resolution, as it sounds, is how things change over time and how quickly you can 53 00:05:13,540 --> 00:05:19,400 do these images together. Your CAT scan, your basic machine in a hospital, 54 00:05:19,400 --> 00:05:24,210 costs 1 to 2.5 million dollars. You probably didn't get one for Christmas 55 00:05:24,210 --> 00:05:30,020 to play around with. Oh well. It's also got this ionizing radiation, you've got 56 00:05:30,020 --> 00:05:32,710 a lot of maintenance, and dedicated technicians. 57 00:05:32,710 --> 00:05:35,240 An MRI, say, your average 3 Tesla magnet 58 00:05:35,240 --> 00:05:41,060 with its own helium quenching chamber no less, as well as dedicated technicians 59 00:05:41,060 --> 00:05:50,710 and experts who can actually read the images. Again $3,000,000. An amazing 60 00:05:50,710 --> 00:05:55,950 and beautiful technology, but really expensive. Amazing spatial resolution, the 61 00:05:55,950 --> 00:06:01,010 best. When it does something at this very high spatial resolution, it actually takes 62 00:06:01,010 --> 00:06:06,490 4 minutes and 16 seconds. Which is a really long time to take to do this 63 00:06:06,490 --> 00:06:11,200 wonderful spatial resolution image. Ultrasound, it's a bit grainy due to 64 00:06:11,200 --> 00:06:18,560 scattering. On average it costs about 1$115k, not too bad. It's a pretty minimal 65 00:06:18,560 --> 00:06:25,460 health risk. EEG. EEG doesn't do any image reconstruction. In fact it does very 66 00:06:25,460 --> 00:06:33,040 little in many ways. But it is still very useful. Your average medical grade by EEG 67 00:06:33,040 --> 00:06:37,920 system is $40k. You might also know of some open source EEG projects which are 68 00:06:37,920 --> 00:06:45,070 pretty cool. So just a note on the radiation of CAT-scans. It's actually the 69 00:06:45,070 --> 00:06:53,700 biggest contributing cause of radiation in the United States. So here I just put 70 00:06:53,700 --> 00:06:58,550 those biomedical imaging modalities onto a graph so that you can kind of think of 71 00:06:58,550 --> 00:07:03,120 them in terms of spatial resolution and time resolution, and where they fall in 72 00:07:03,120 --> 00:07:09,240 the picture of common things that go wrong with people. Like, X-rays or CAT scans are 73 00:07:09,240 --> 00:07:15,430 great for for looking at bone and bone breaks; pulmonary edema, that's water on 74 00:07:15,430 --> 00:07:20,760 the lung ,tuberculosis, huge in third- world countries, massive problem. You 75 00:07:20,760 --> 00:07:25,380 don't actually need super high spatial resolution to be able to detect it. And 76 00:07:25,380 --> 00:07:29,610 it's important to sort of understand what you can do at different spatial and time 77 00:07:29,610 --> 00:07:35,430 resolutions. Under like, the optimal goal of all of this, I put non-invasive 78 00:07:35,430 --> 00:07:40,370 electrophysiology. What that is, is high spatial resolution and high time 79 00:07:40,370 --> 00:07:46,820 resolution. That's where you can measure ion activation, or basically what cells 80 00:07:46,820 --> 00:07:51,460 are doing when they communicate with each other, which is right now only done in an 81 00:07:51,460 --> 00:07:55,850 invasive manner. Today I'm gonna talk about this new 82 00:07:55,850 --> 00:08:00,870 technique called Electrical Impedance Tomography and describe where it will fit 83 00:08:00,870 --> 00:08:09,650 in amongst what already exists. So what is it. Okay yeah basically you send AC 84 00:08:09,650 --> 00:08:17,130 currents through the body, say a 50 kilohertz current. And that will take 85 00:08:17,130 --> 00:08:23,360 different routes based on what tissue there is. So it might go around some cells 86 00:08:23,360 --> 00:08:29,620 and straight through others. And that's really important because differentiating, 87 00:08:29,620 --> 00:08:34,869 say, fat from muscle is one thing that you could do. But you can go further and 88 00:08:34,869 --> 00:08:41,969 differentiate, say, tumors from healthy tissue. Because tumors have different 89 00:08:41,969 --> 00:08:47,960 impedance spectra to the healthy tissue. So as you can see, that would be very 90 00:08:47,960 --> 00:08:53,470 useful to do. This set up here is a called a phantom. What it is, it's like a 91 00:08:53,470 --> 00:08:58,650 simulated human body. You get some saltwater - the body is 80% water as you 92 00:08:58,650 --> 00:09:03,369 might know -you get some meat or vegetables. You put it inside and then you 93 00:09:03,369 --> 00:09:07,249 use that to image. So we have current flowing through all these different 94 00:09:07,249 --> 00:09:13,120 directions and we recreate an image. Right now it's used for lung volume 95 00:09:13,120 --> 00:09:18,399 measurements. This is a baby with an EIT setup. Muscle and fat mass, there's a 96 00:09:18,399 --> 00:09:22,180 paper on gestural recognition that just came out this year, you can look at 97 00:09:22,180 --> 00:09:27,260 bladder and stomach fullness. There's some research papers on breast and kidney 98 00:09:27,260 --> 00:09:33,860 cancer detection. There's another research paper on hemorrhage detection for stroke. 99 00:09:33,860 --> 00:09:39,110 You can also look at the ... there's more R&D on the depth of anesthesia in in 100 00:09:39,110 --> 00:09:42,820 surgery as well, which would be another interesting use for it. So all of these 101 00:09:42,820 --> 00:09:50,649 are sort of in the works and you might ask, "Great, that sounds amazing, why 102 00:09:50,649 --> 00:09:55,589 isn't everybody using it already?" Well yeah it's really an R&D technique right 103 00:09:55,589 --> 00:10:03,029 now and it has a big problem: its spatial resolution seems pretty limited. So it's 104 00:10:03,029 --> 00:10:06,990 limited by the number of electrodes. But I will discuss some potential ways to get 105 00:10:06,990 --> 00:10:12,929 around that. As we go, it might not ever get to the spatial resolution of MRI. 106 00:10:12,929 --> 00:10:16,740 But maybe we don't need it to to be useful. Because it's so compact. It's so 107 00:10:16,740 --> 00:10:23,810 cheap, nothing about it is expensive. It's got better source localization than EEG. 108 00:10:23,810 --> 00:10:27,759 It does not ionize, it's not harmful to human tissue. It's 109 00:10:27,759 --> 00:10:33,589 also got great time resolution, so it has advantages and disadvantages. I'll just 110 00:10:33,589 --> 00:10:39,240 remind you of what the first MRI scan looked like at this point in time. As you 111 00:10:39,240 --> 00:10:45,599 can see it looks pretty crappy in 1977. And now it looks pretty awesome. That's a 112 00:10:45,599 --> 00:10:51,230 slice of my head by the way in a 3 Tesla MRI scanner. This is what early EIT looks 113 00:10:51,230 --> 00:10:59,429 like. That's with 16 electrodes only. What will it look like in a few years time I 114 00:10:59,429 --> 00:11:06,899 don't know. I hope that MRI gives you a pathway that it will take take too. 115 00:11:06,899 --> 00:11:13,110 Now I'll introduce you to the OpenEIT project. The OpenEIT project is obviously 116 00:11:13,110 --> 00:11:20,449 open source. It has a PCB design done in Eagle CAD. It has firmware written in C. 117 00:11:20,449 --> 00:11:24,919 It has a Python dashboard that lets you see the reconstruction in real time. It 118 00:11:24,919 --> 00:11:29,300 also has a reconstruction algorithm which I'll go into. And you can get it from 119 00:11:29,300 --> 00:11:36,540 github right there. So how does it reconstruct an image? OpenEIT right now 120 00:11:36,540 --> 00:11:43,059 has 8 electrodes and what you do is, you send this 50 kHz current through every 121 00:11:43,059 --> 00:11:48,920 combination of those 8 electrodes and you get a different impedance value for each 122 00:11:48,920 --> 00:11:55,519 of those measurements. On the left you can see basically what you're doing. You know where the 123 00:11:55,519 --> 00:12:01,269 electrodes are positioned and you get one value going horizontally. You add it to 124 00:12:01,269 --> 00:12:06,099 another value coming from another direction. And again, you can sort of see 125 00:12:06,099 --> 00:12:11,240 it's getting a low resolution image as it goes around adding those values together. 126 00:12:11,240 --> 00:12:18,370 If you use many, many views you bring the image back. This is the radon transform, 127 00:12:18,370 --> 00:12:23,749 that's what it's called, and you basically just send lots of current 128 00:12:23,749 --> 00:12:26,900 through these different slightly different angles and you build up something called a 129 00:12:26,900 --> 00:12:33,580 sinogram which is over there. And then you invert it to get the image back. I used 130 00:12:33,580 --> 00:12:37,209 OpenCV which is a really common image processing library to do this. You can 131 00:12:37,209 --> 00:12:43,360 just do it with a regular image yourself and try it out. But what I did is exactly 132 00:12:43,360 --> 00:12:49,069 the same as what you do with a regular image, except I use current to be the 133 00:12:49,069 --> 00:12:57,810 input data. So this is the PCB design in Eagle. Basically it has a 134 00:12:57,810 --> 00:13:02,830 few different features. A connector for your 8 electrodes. It's running an ARM 135 00:13:02,830 --> 00:13:11,339 Cortex M3, which is quite nice. It has a dedicated DFT engine for doing your direct 136 00:13:11,339 --> 00:13:16,829 Fourier transform in real, time which is also quite nice. A JTAG debugger to easily 137 00:13:16,829 --> 00:13:22,619 reprogram it. It's got coin cell or external battery options. It has UART to 138 00:13:22,619 --> 00:13:28,660 get the serial data off. And you can also flip it to Bluetooth mode and get the data 139 00:13:28,660 --> 00:13:32,040 off by Bluetooth if you felt like going Wireless. 140 00:13:32,040 --> 00:13:37,131 At this point you might be asking "Is this safe for me to play around with?", which 141 00:13:37,131 --> 00:13:42,879 is a really great question because the answer is actually "Yeah! it is". There's 142 00:13:42,879 --> 00:13:51,050 some guidelines called the IEC60601-1 guidelines for safer use in humans. And 143 00:13:51,050 --> 00:13:56,949 basically which says it should be, and openEIT is less than 10 micro amps which 144 00:13:56,949 --> 00:14:02,959 is great because that's well within their guidelines. If you want to compare it to 145 00:14:02,959 --> 00:14:06,220 other things that are completely legal, say I don't know if you've seen there's 146 00:14:06,220 --> 00:14:10,629 like late-night TV ads for those abs stimulators that stimulate your muscles, 147 00:14:10,629 --> 00:14:17,269 there are about 15 to 20 milliamps just for reference and as a scale to look at 148 00:14:17,269 --> 00:14:23,749 the 10 micro amps. So some of you might have used them already and that's hugely 149 00:14:23,749 --> 00:14:27,890 more current than what we're putting through to image the body here. This is 150 00:14:27,890 --> 00:14:33,480 what the dashboard looks like. It does the reconstruction. You can connect to serial 151 00:14:33,480 --> 00:14:37,850 at baseline. You can obviously adjust sliders to look at the area that you want 152 00:14:37,850 --> 00:14:43,849 to look at. You can read from a file and fiddle around however you would like to. 153 00:14:43,849 --> 00:14:49,169 This is what it looks like when you reconstruct something. I have a phantom up 154 00:14:49,169 --> 00:14:54,079 there which is a part of water with a cup in it. I moved the cup around anti- 155 00:14:54,079 --> 00:14:58,900 clockwise so you can see in each of the pictures I move it around a little bit 156 00:14:58,900 --> 00:15:04,400 more. And you can see the reconstruction there with me moving the cup around again. 157 00:15:04,400 --> 00:15:07,969 This might not be wow-ing you with the resolution, with only 8 electrodes. It's a 158 00:15:07,969 --> 00:15:14,800 proof of concept but that's okay. Let's see if we can make this I make this go. 159 00:15:14,800 --> 00:15:20,720 Here's a real-time video demonstration of it. Here's me with a shot glass. I'm 160 00:15:20,720 --> 00:15:24,639 moving around anti-clockwise. Hopefully you can see on the left the image being 161 00:15:24,639 --> 00:15:33,920 reconstructed in real time. And there we go, move to the bottom. You can see it 162 00:15:33,920 --> 00:15:41,030 over there and again up to the top. you can see it over there. So that's a basic 163 00:15:41,030 --> 00:15:44,739 proof of principle version of it running. 164 00:15:49,129 --> 00:15:55,079 So the first MRI scan of human lungs wasn't that amazing. 165 00:15:55,079 --> 00:15:57,634 Early EIT scan wasn't either. 166 00:15:57,634 --> 00:16:02,824 *applause* Something else that you can use it 167 00:16:02,824 --> 00:16:08,590 that for is differentiating objects. Multi-frequency. This is what they're 168 00:16:08,590 --> 00:16:13,969 doing the breast cancer and kidney cancer scans on. Basically you send different 169 00:16:13,969 --> 00:16:17,920 frequencies through these times, called multi-frequency Electrical Impedance 170 00:16:17,920 --> 00:16:23,180 Tomography and you build up a spectrum. Here I've got an apple, a pear oh no a 171 00:16:23,180 --> 00:16:27,660 sweet potato and and some water. And I've sent through these different frequencies 172 00:16:27,660 --> 00:16:32,309 and I get these different spectrums. They're different, you can see that 173 00:16:32,309 --> 00:16:35,829 they're different. They're quite obviously different but yeah you can also just 174 00:16:35,829 --> 00:16:39,680 simply classify. And on the left you can see where the water is, the apple is, the 175 00:16:39,680 --> 00:16:44,769 sweet potato is. Or, the sweet potato and the apple a little bit harder that one. 176 00:16:44,769 --> 00:16:53,800 But that's basically what you do when you detect cancer. So that's what I did. But 177 00:16:53,800 --> 00:16:57,760 maybe we should look at the other papers and see what they did because they did 178 00:16:57,760 --> 00:17:04,589 better than me. So there's this guy called Aristovich, 2014 he published spatial and 179 00:17:04,589 --> 00:17:07,569 temporal resolution, and using this technique 200 micro meters less than 2 180 00:17:07,569 --> 00:17:14,230 milliseconds which covers most of the applications that I listed on that graph 181 00:17:14,230 --> 00:17:19,260 at the start of the talk. The downside here is that it was an intracranial array, 182 00:17:19,260 --> 00:17:24,190 so it was under the skull. So very dense electrodes, a lot more electrodes. I only 183 00:17:24,190 --> 00:17:32,089 used 8 he used like 256 so you can see that it can be, like, the potential is 184 00:17:32,089 --> 00:17:36,859 there. So how should we use it first? what's a 185 00:17:36,859 --> 00:17:41,230 nice low hanging through fruit? What about medical imaging in the developing world 186 00:17:41,230 --> 00:17:46,649 where I believe 4 billion people don't have access to medical imaging. No MRI, no 187 00:17:46,649 --> 00:17:51,519 CAT scans. Why is the EIT good for that? It's cheap to mass-produce, super 188 00:17:51,519 --> 00:17:58,089 portable, super low power. So that would be a great place to start. What could we 189 00:17:58,089 --> 00:18:05,330 do first? I'm going to go back to this image again and have a look. Tuberculosis 190 00:18:05,330 --> 00:18:09,110 affects a lot of people in the developing world and you don't need amazing spatial 191 00:18:09,110 --> 00:18:14,901 resolution to detect it. That would be a good one. Or what about a pulmonary edema? 192 00:18:14,901 --> 00:18:21,660 Pulmonary edema is water on the lung. It's actually already used for that. You can 193 00:18:21,660 --> 00:18:26,640 quite easily see the different volume present, or the different conductivity 194 00:18:26,640 --> 00:18:33,590 maps it's called, of a working lung and a not so working lung right there. 195 00:18:33,590 --> 00:18:41,410 Next steps. So what should we do to make this technique better? What should we do 196 00:18:41,410 --> 00:18:48,200 for OpenEIT to make it better? If you want to innovate again, that's the github 197 00:18:48,200 --> 00:18:53,460 project. Just go ahead. Oh that's an avocado, it has a seat in the middle. Who 198 00:18:53,460 --> 00:19:05,519 knew? I do. So I see the two main routes forward as: One would be this low-cost biomedical 199 00:19:05,519 --> 00:19:10,690 imaging for the developing world. You could just stick with the static imaging 200 00:19:10,690 --> 00:19:16,270 reconstruction because why not. you'd need a few more electrodes than it currently 201 00:19:16,270 --> 00:19:22,240 has. One of the main problems with the technique is how you stick it to the skin. 202 00:19:22,240 --> 00:19:25,730 So my suggestion for that is why don't you just use a water bath and stick the body 203 00:19:25,730 --> 00:19:31,929 part of interest in a body of water, because water gets rid of a lot of the, 204 00:19:31,929 --> 00:19:38,390 it's called the contact impedance problem. Or, on the kind of exciting science front, 205 00:19:38,390 --> 00:19:47,230 you've got the advancing neuroscience option. Which would be measuring both high 206 00:19:47,230 --> 00:19:50,340 spatial resolution and high time resolution. So that's the non-invasive 207 00:19:50,340 --> 00:19:57,710 electrophysiology solution. Or, and that would be super awesome, there's a couple 208 00:19:57,710 --> 00:20:03,629 of ways forward to do that and I'm going to sort of discuss each of those. 209 00:20:03,629 --> 00:20:10,169 So roughly there's physical configuration improvements that could be done. There's 210 00:20:10,169 --> 00:20:14,480 things that you can do to improve the spatial resolution. There's things you can 211 00:20:14,480 --> 00:20:19,480 do to improve the time resolution. And this is interesting tack on at the end 212 00:20:19,480 --> 00:20:25,210 that I thought I'd mentioned, which is 'write' functionality. So we're using very 213 00:20:25,210 --> 00:20:33,090 small currents to read an image. What if we pumped the current up a little before 214 00:20:33,090 --> 00:20:38,939 you know it you're writing. I think not invasive deep brain stimulation in a 215 00:20:38,939 --> 00:20:48,670 focused way, that would be very very cool. So, contact impedance. Major problem right 216 00:20:48,670 --> 00:20:54,059 now, there is a well-known solution I haven't done it yet you do this thing 217 00:20:54,059 --> 00:21:00,909 called differential referencing, common mode rejection should be done I haven't 218 00:21:00,909 --> 00:21:05,069 done it that's the next step. That means that it will work when you just attach it 219 00:21:05,069 --> 00:21:10,630 with electrodes on the body. What happens is, electrodes have a like some 220 00:21:10,630 --> 00:21:16,620 capacitance and different amounts which kind of interfere with the the measurement 221 00:21:16,620 --> 00:21:19,690 that you want to make which you want to be very accurate and just of your body. You 222 00:21:19,690 --> 00:21:24,529 don't want to include the electrode information in there that's changing. 223 00:21:24,529 --> 00:21:30,110 There's a way to remove that that's well known already. Another physical 224 00:21:30,110 --> 00:21:34,870 configuration improvements: just increase the number of electrodes. Wonderful, now 225 00:21:34,870 --> 00:21:41,640 you've just improved the resolution. Or the placing the part in water. Another set 226 00:21:41,640 --> 00:21:46,700 of next steps would be on the mathematical side. I mentioned that I use linear back 227 00:21:46,700 --> 00:21:55,759 projection which is a wonderful technique, that's how they do CAT scans. With X-rays 228 00:21:55,759 --> 00:21:59,310 that's exactly what they do. However, it makes some appalling 229 00:21:59,310 --> 00:22:06,649 assumptions, like parent moves and straight lines. That is not true. What you 230 00:22:06,649 --> 00:22:11,209 should do is get a finite element model and solve Maxwell's equations because 231 00:22:11,209 --> 00:22:18,630 current bends around objects. Actually it works in three dimensions too which might 232 00:22:18,630 --> 00:22:23,639 not be all that surprising but it needs to be solved for those three dimensions which 233 00:22:23,639 --> 00:22:26,810 is why you just need to solve Maxwell's equations and 234 00:22:26,810 --> 00:22:30,899 create a finite element model. And there's a quite a bit of work on 235 00:22:30,899 --> 00:22:34,610 mathematical solutions that get higher resolution. 236 00:22:34,610 --> 00:22:42,189 That's another improvement area. And now as I mentioned this awesome new technique. 237 00:22:42,189 --> 00:22:44,649 Which, actualy there's a paper on this year called 238 00:22:44,649 --> 00:22:50,990 magneto-acoustic electical tomography. You might remember 239 00:22:50,990 --> 00:22:55,580 the FBI rule from high school. When you have a current flowing, 240 00:22:55,580 --> 00:23:02,429 perpendicular to that there will be a force. Now that force, say it's vibrating 241 00:23:02,429 --> 00:23:07,309 with 50 kilohertz. that's the AC signal that you're sending through. Now you have 242 00:23:07,309 --> 00:23:11,460 a vibrating compression wave. That's sound. You can pick that up with a little 243 00:23:11,460 --> 00:23:19,580 piezoelectric element. And that's actually a focus of work. From that you can get 244 00:23:19,580 --> 00:23:27,070 really good edge information, because as I mentioned earlier, sound scatters at 245 00:23:27,070 --> 00:23:31,460 edges. So you would also get the electrical impedance tomography 246 00:23:31,460 --> 00:23:38,549 information for the tissue sensitivity. Why not combine those results together and 247 00:23:38,549 --> 00:23:43,259 you would have a better tool. It currently gets lesser resolution in the middle 248 00:23:43,259 --> 00:23:50,180 simply from how you every combination of electrodes just ends up having a less 249 00:23:50,180 --> 00:23:56,799 dense number in the middle. You can also do something as simple as increasing the 250 00:23:56,799 --> 00:24:02,049 power that you send through if you're game to do that. This is a kind of gory 251 00:24:02,049 --> 00:24:07,961 picture. Right now epileptics, if they're really troubled by their problem, which 252 00:24:07,961 --> 00:24:13,460 they are often, they go into a hospital have their brains opened up and they 253 00:24:13,460 --> 00:24:18,580 stick this array on their head through their skull. And they leave it open 254 00:24:18,580 --> 00:24:24,340 for a week. And they try to induce seizures through sleep deprivation. 255 00:24:24,340 --> 00:24:30,440 And then they measure the activation potentials that way to locate the foci or 256 00:24:30,440 --> 00:24:36,200 where they going to do surgery to stop you from having seizures. But it would be much 257 00:24:36,200 --> 00:24:40,260 better and nicer if you could do it not invasively and you probably can if you 258 00:24:40,260 --> 00:24:43,600 improve the time resolution of EIT. there's nothing stopping you from doing 259 00:24:43,600 --> 00:24:50,360 that by the way. You just have to, like, it's just a next step really. 260 00:24:50,360 --> 00:24:56,919 And then I'll also mention write- functionality. So there was a paper that 261 00:24:56,919 --> 00:25:02,700 came out halfway through this year by a guy called Neil Grossman (?) and what he 262 00:25:02,700 --> 00:25:09,170 did is, he showed that you can stimulate neurons by sending current through the 263 00:25:09,170 --> 00:25:18,739 skull and in a focused way. Now why that's interesting is, you can non-invasively 264 00:25:18,739 --> 00:25:23,190 stimulate neurons. So that's the write- functionality. It's unknown what 265 00:25:23,190 --> 00:25:27,950 resolution is or how well you could control the the focal point here. But it 266 00:25:27,950 --> 00:25:34,220 works in the principle of beat frequencies so he sent through two kilohertz and 2.05 267 00:25:34,220 --> 00:25:42,379 kilohertz and basically had a beat frequency of 10 Hertz arise from that and 268 00:25:42,379 --> 00:25:50,279 basically stimulated neurons in this area that he can control via an x- and y-axis 269 00:25:50,279 --> 00:25:59,940 which is very impressive. Leaves a lot of questions open. Those are some possible 270 00:25:59,940 --> 00:26:06,309 next steps that it could go in. Obviously I think this is interesting. I hope that 271 00:26:06,309 --> 00:26:11,340 you do too. I'd love it if you would want to sign up to a mailing list I'll give a 272 00:26:11,340 --> 00:26:17,049 link on the next page. If you want to collaborate email me. If you know any 273 00:26:17,049 --> 00:26:22,070 funding bodies that might be interested in the developing medical imaging for 274 00:26:22,070 --> 00:26:26,330 the third world I'd love to be put in contact. If you wanted a kit and, if there 275 00:26:26,330 --> 00:26:30,230 were enough people that wanted a kit, probably of the next version which would 276 00:26:30,230 --> 00:26:36,200 have 32 electrodes sign up to the mailing list, talk to me. Thanks. 277 00:26:36,200 --> 00:26:46,820 *applause* Rintoul: Thank you 278 00:26:46,820 --> 00:26:49,690 *applause* Herald Angel: Thank you very much. We have a little bit 279 00:26:49,690 --> 00:26:57,610 time for Q&A. And please if you have to leave the room make it in a very quiet 280 00:26:57,610 --> 00:27:05,730 way. So is there ... there are some questions I've seen microphone 4 first. 281 00:27:05,730 --> 00:27:08,970 Please go ahead. Audience member: So, a great thing 282 00:27:08,970 --> 00:27:15,649 thinking about developing countries and getting them medical tech. But at the very 283 00:27:15,649 --> 00:27:20,889 first beginning you said imagine a world where this imaging would be all available 284 00:27:20,889 --> 00:27:26,610 like every day and it creeped me out a little bit. Do you really think that it's 285 00:27:26,610 --> 00:27:33,740 a good idea to go in the shower in the morning and have your I don't know your 286 00:27:33,740 --> 00:27:40,019 bathtub telling you that there is a small mass inside your lungs. 287 00:27:40,019 --> 00:27:46,889 Rintoul: That's a good question. Basically the question was: There's a privacy 288 00:27:46,889 --> 00:27:52,360 concern with looking inside your body. It doesn't sound that great to some people. 289 00:27:52,360 --> 00:27:56,200 To those people I would say you should turn off I know that sounds a little 290 00:27:56,200 --> 00:28:05,179 harsh. But please just turn it off, don't use it. And with all scientific movements 291 00:28:05,179 --> 00:28:12,080 forward comes great risk, I also say. And it can be used for good or evil and it's 292 00:28:12,080 --> 00:28:17,769 up to us as a society how we want to choose to use it. And how we structure 293 00:28:17,769 --> 00:28:24,510 ourselves and potentially motivate and incentivize corporations to use it in a 294 00:28:24,510 --> 00:28:31,919 responsible way. Part of making this open is I hope that, basically if people have 295 00:28:31,919 --> 00:28:36,470 access to it you can choose for yourself how you'd want to use it. 296 00:28:36,470 --> 00:28:40,809 Herald Angel: And next question would be from the Signal Angel please. 297 00:28:40,809 --> 00:28:45,070 Signal Angel: Yes I have a couple of questions from the internet. First of all, 298 00:28:45,070 --> 00:28:51,439 what type of AC frequencies in use? the asker assumes sinusoidal but he wonders if 299 00:28:51,439 --> 00:28:54,779 you also tried square wave, triangular and other shapes. 300 00:28:54,779 --> 00:29:00,360 Rintoul: That's also a really interesting question. It's about what kinds of waves 301 00:29:00,360 --> 00:29:08,580 are used, what kinds of AC signals. Typically it's done with AC sine waves 302 00:29:08,580 --> 00:29:14,500 ranging all over the place, depending on what application you want to use up for. I 303 00:29:14,500 --> 00:29:19,980 mentioned multi frequency EIT for cancer detection. That uses a lot of different 304 00:29:19,980 --> 00:29:26,090 frequencies so if you wanted to use other waveforms I think that would be really 305 00:29:26,090 --> 00:29:32,840 interesting. Nobody's tried, you can, that should be done. 306 00:29:32,840 --> 00:29:38,740 Herald: So since there's a big queue on microphone 3 I would go there please. 307 00:29:38,740 --> 00:29:44,630 Audience member: Yes I have a technical question. Assuming that you won't use this 308 00:29:44,630 --> 00:29:50,590 techniques on humans or organic matter at all and what are the limitations for the 309 00:29:50,590 --> 00:29:56,190 resolution. The spatial resolution. And is there a possibility to reduce the spatial 310 00:29:56,190 --> 00:29:59,100 resolution. Rintoul: You mean increase the spatial 311 00:29:59,100 --> 00:30:06,179 resolution or reduce it? Audience member: Reduce the voxel size 312 00:30:06,179 --> 00:30:12,470 Rintoul: So increase the spatial resolution. Yes absolutely. So I was 313 00:30:12,470 --> 00:30:16,240 trying to go through a few of the next steps that could get to that. One of them 314 00:30:16,240 --> 00:30:21,259 is magneto-acousto electrical tomography because you get two different types of 315 00:30:21,259 --> 00:30:27,580 information which you could put together to form a higher resolution image. So 316 00:30:27,580 --> 00:30:33,320 that's one way and if you didn't need to worry about human safety I recommend you 317 00:30:33,320 --> 00:30:39,240 just turn the power up, that will also work. 318 00:30:39,240 --> 00:30:46,389 Herald: Okay I think we go back to the signal angel for one short one please. 319 00:30:46,389 --> 00:30:49,890 Signal Angel: Yes I have another question from the internet. from a doctor this 320 00:30:49,890 --> 00:30:54,340 time. He wonders if there are any clinical studies that compare pulmonary edema 321 00:30:54,340 --> 00:30:59,919 diagnostics with EIT to ultrasound and why don't we just work on cheap ultrasound 322 00:30:59,919 --> 00:31:03,139 instead. Rintoul: That's a good question. People 323 00:31:03,139 --> 00:31:08,289 are working on cheap ultrasounds. Ultrasound gives different information to 324 00:31:08,289 --> 00:31:14,360 EIT. It has a problem of the sound scattering. So it's a different type of 325 00:31:14,360 --> 00:31:21,289 information which has different pros and cons. And and I think people should make 326 00:31:21,289 --> 00:31:27,169 cheap ultrasound. And I would like to see the hybrid modality come together. You can 327 00:31:27,169 --> 00:31:31,769 get really good tissue distinction with EIT so there's pros and cons. 328 00:31:31,769 --> 00:31:36,680 Herald: Okay then, microphone 2 please. Audience member: You had a really good 329 00:31:36,680 --> 00:31:44,649 talk my question so far you always need direct contact to the electrode, right? So 330 00:31:44,649 --> 00:31:50,519 it has to be direct contact or in water. Is there way to detect or measure the 331 00:31:50,519 --> 00:31:56,659 signal without direct contact? So maybe in if the if the object is in air or any 332 00:31:56,659 --> 00:32:01,419 other gas? Rintoul: Right. I wish there was. No is 333 00:32:01,419 --> 00:32:07,570 the short answer. Unless ... Audience member: Any research on making it 334 00:32:07,570 --> 00:32:11,129 happen? Rintoul: Well yeah you can you can use 335 00:32:11,129 --> 00:32:20,129 X-rays. They work wonderfully to to go through the air. But if you use them I 336 00:32:20,129 --> 00:32:24,220 mean you do increase your chance of cancer so don't use them all the time on 337 00:32:24,220 --> 00:32:29,499 yourself. Again CAT scanners are a little bit expensive. 338 00:32:29,499 --> 00:32:35,529 Herald: Thank you and I think we have time for one more from microphone 3 339 00:32:35,529 --> 00:32:41,970 Audience member: My question would be what, so maybe I've missed it, but what's 340 00:32:41,970 --> 00:32:47,009 the order of magnitude for cost so would this be feasible at like a hackerspace for 341 00:32:47,009 --> 00:32:53,749 this to implement. And does the industry see the possibility to make money. 342 00:32:53,749 --> 00:33:00,779 Rintoul: Yes a lot of those sort of these early like R&D papers yeah they should be 343 00:33:00,779 --> 00:33:06,870 applied and you could make money with it absolutely. And there's no component in 344 00:33:06,870 --> 00:33:14,960 there that costs more than a couple of cents. I suppose a cortex m3 like costs a 345 00:33:14,960 --> 00:33:19,570 couple of dollars. And I mean I don't know what your budget is but yes you I think 346 00:33:19,570 --> 00:33:24,309 you could do this in a hackerspace without any problems. There's nothing stopping 347 00:33:24,309 --> 00:33:29,590 anyone from doing this and as we know microcontrollers are becoming cheaper and 348 00:33:29,590 --> 00:33:36,460 cheaper. So why not. Herald: I don't get Hasty's signs from the 349 00:33:36,460 --> 00:33:40,059 sideline so I think I can take another question from 2 please. 350 00:33:40,059 --> 00:33:46,600 Audience member: So far you have showed us images of 2d planes. What about volumes 351 00:33:46,600 --> 00:33:52,369 Rintoul: Yes so there's work on solving for volumes using finite element models 352 00:33:52,369 --> 00:34:02,759 and solving Maxwell's equations. Basically I just did the shortest route to reach 353 00:34:02,759 --> 00:34:07,670 image reconstruction that was available which was linear back projection which is 354 00:34:07,670 --> 00:34:12,280 typically done in a 2d plane. So absolutely, you can do it in three 355 00:34:12,280 --> 00:34:16,370 dimensions. Herald: So I'm very sorry we are out of 356 00:34:16,370 --> 00:34:23,550 time the queue back there you can have the chance to chat with our speaker just right 357 00:34:23,550 --> 00:34:31,880 now. The next talk coming up is in about 15 minutes and it's I think also in 358 00:34:31,880 --> 00:34:36,920 English. See you then and a big round of applause for our speaker, excuse me. 359 00:34:36,920 --> 00:34:42,460 *applause* 360 00:34:42,460 --> 00:34:59,990 *music*