WEBVTT 00:00.000 --> 00:12.700 Hello everybody, my name is Jack Twitty and today I'm going to be getting an introduction to the world of neuroengineering, how it intersects with biohacking, and how you might one day grow a brain in a jar yourself. 00:13.540 --> 00:16.620 First some real quick biotype information about myself. 00:16.620 --> 00:24.100 I'm currently a doctoral student in the UNC NCSU Joint Department of Biomedical Engineering in the Research Triangle part of North Carolina. 00:24.340 --> 00:31.180 You may or may not have heard of one or both of those schools depending on whether or not you're a basketball fan, so I've included a map just to be safe. 00:32.160 --> 00:39.300 So even though I've been working in STEM for a while now, I actually started out as a philosophy and studio art major in undergrad at UNC. 00:39.420 --> 00:46.620 I always like to bring up that disclaimer because I feel it's important to, you know, destroy any bit of credibility I might have early on in the talk like this. 00:46.960 --> 01:00.840 After undergrad I took a hard right turn into engineering and after teaching myself some electronics for fun and getting myself involved in 3D printing, which is sort of the hacker gateway drug, I decided I liked that kind of thing and I wanted to pursue it full-time. 01:01.760 --> 01:13.800 So I went on to get an engineering master's at NC State focusing on microfabrication and then I moved into biomedical engineering specifically first at the City College of New York and then back at a joint department run by both of my earlier schools. 01:14.900 --> 01:25.020 Another thing I'm interested in is implantable hardware development and I'm actually in collaboration with some of the other Biohacking Village speakers on a few projects in that area, so stay tuned next year for that hopefully. 01:25.500 --> 01:31.060 And in the almost non-existent spare time I have outside of all that, I also work as an EMT occasionally. 01:32.120 --> 01:40.860 So this talk is not going to be focused on my own research but just to cover that briefly, my main interest is in the development of tools for neuroscience and neurology research. 01:40.860 --> 01:47.040 I'm really interested in studies of consciousness in particular as sort of a holdover from my past life as a philosophy major. 01:47.980 --> 02:02.940 Some specific areas I've worked in in the include electronics and hardware for a variety of neurostimulation devices, circuits for energy harvesting and self-powered systems, a variety of medical type sensing devices, primarily in the form of wearables for health care. 02:03.920 --> 02:12.160 My current research focus involves using electrochemistry to detect and monitor neurotransmitters both in vivo and in culture models. 02:12.320 --> 02:21.560 And for my doctoral work I'm developing a cell culture system to try and explore some questions related to the simultaneous release of two neurotransmitters in a specific region of the brain. 02:23.260 --> 02:33.940 So again, my goal with this talk is not really to cover my own research in this area but rather to provide a very , very high-level overview of neuroengineering and some interesting work that falls under that umbrella . 02:34.340 --> 02:41.380 So neuroengineering is very loosely defined and even a very restrictive slice of that kind of work would still be a tremendous amount of ground to cover. 02:41.480 --> 02:47.080 Since it's an impossible task to try to do this field justice in such a short talk, I'm not really going to try to do that. 02:47.280 --> 02:55.760 Instead I'll be providing a roadmap for the kind of work being done in neuroengineering and I'll be highlighting some particular developments that I feel are important just to give a sampling. 02:56.220 --> 03:07.560 This is intended merely as a starting point and in that spirit I'll conclude by discussing what might be involved in getting started in neuroengineering in a garage, basement workshop, or other improvised mad science lab if you're choosing. 03:09.260 --> 03:18.200 So to start out, because neuroengineering is such an expansive and diverse field, I think it's useful to start with some basic delineations of the kind of work being done in that space. 03:18.600 --> 03:26.200 Many different groups define neuroengineering in different ways, sometimes vastly so, and that can make categorizing this type of work fairly difficult. 03:26.540 --> 03:31.820 And since all these different groups are putting forth their own definitions of the term, I figure I might as well join in. 03:31.820 --> 03:38.420 So I'm going to share my own totally arbitrary ad hoc taxonomy of neuroengineering as it makes sense to me. 03:38.660 --> 03:39.920 Your mileage may vary. 03:40.340 --> 03:46.780 I like to divide up neuroengineering projects based on where the system is being used and specifically what it's being used for. 03:47.060 --> 03:57.740 The first main branch under this scheme is in vivo neuroengineering and that's probably the part of neuroengineering people in general are most familiar with and what you might think of when I say the term neuroengineering. 03:58.200 --> 04:05.480 I define this branch to include systems that involve creating an interface with the nervous system of a living human or other animal if applicable. 04:05.820 --> 04:14.000 This includes a huge range of equipment encompassing both invasive and non-invasive devices which can be used for sensing stimulation or in some cases both. 04:14.000 --> 04:31.980 This family includes things like prosthetic devices that interact with peripheral nerves, stimulators found anywhere in the body like pacemakers, deep brain stimulation leads, EEG arrays, electrocorticogram probes, and many other devices depending on where exactly you want to draw the line between nervous system and not nervous system. 04:32.260 --> 04:41.720 As far as recent development goes, this is the category into which virtually all brain interfaces would fall which have been a huge area of active research for the past few decades. 04:43.000 --> 04:51.740 The second main branch of neuroengineering is the sort of the odd one out and I think a lot of people might not actually categorize it as neuroengineering proper. 04:52.140 --> 04:56.660 But that being said, I think its potential overlap with the other two branches justifies its inclusion. 04:56.880 --> 05:07.620 If we think of in vivo neuroengineering as using human-designed electronic systems to work in a biological context, neuromorphic engineering could be posed as the exact opposite of this process. 05:07.720 --> 05:19.180 So in neuromorphic engineering, researchers take inspiration from the behavior of the nervous system and apply those behaviors and mechanisms to problems of computation, often in a context that doesn't directly involve biology at all. 05:19.420 --> 05:31.300 This works because the nervous system of humans and other animals exists primarily as a powerful processing and control system, one which has evolved over millions of years to be highly robust in the face of interference and edge cases. 05:31.300 --> 05:36.000 These are things that computing systems, we typically envision them, are notoriously bad at handling. 05:36.540 --> 05:49.340 So while a nervous system can't compete with digital circuits in terms of processing volume and speed, all its being equal, many real-world environments would benefit tremendously from the kind of flexibility offered by neuro-inspired processing. 05:49.680 --> 05:59.760 In particular, this combined analog and digital nature of the sum and fire model of neural processing, where a bunch of digital inputs are weighted in an analog fashion and determine a single digital output. 06:00.020 --> 06:04.300 This allows for advanced categorization and even quasi-learning behaviors. 06:04.680 --> 06:12.120 As testament to the applicability of this idea, just look at the explosion of the use of neural networks as part of the machine learning tool set in recent years. 06:12.260 --> 06:20.580 While these are implemented almost exclusively in software, the same neuro-inspired principles apply in a loose way to vastly improve performance in specific use cases. 06:20.800 --> 06:27.040 And under this totally arbitrary taxonomy, neural networks would also fall under the loose umbrella of neuroengineering. 06:27.280 --> 06:32.300 However, more interesting, in my opinion, are recent attempts to apply the same strategy to hardware directly. 06:32.640 --> 06:42.140 So in this approach, electronic stand-ins for neurons, which vary in terms of how closely they mimic actual neural behavior, are assembled together to create a processing unit as needed. 06:42.140 --> 06:54.880 So this often involves either the use of microelectronic circuits consisting of multiple conventional components, which together generate neuron-like behavior, or in more recent examples, the use of newer exotic components to directly create this desired behavior. 06:54.880 --> 07:08.040 So by designing integrated circuits to intrinsically operate in a neuro-inspired fashion rather than implementing this behavior in software using a conventional digital processor, this process can be massively paralyzed and sped up, greatly increasing the benefits of this approach. 07:08.180 --> 07:12.400 The final branch in this taxonomy is one that's probably less well-known than the other two. 07:12.400 --> 07:21.280 In in vitro neuroengineering, neural cells themselves are used in the construction of other systems engineered towards some purpose other than their intrinsic biological function. 07:21.560 --> 07:31.640 So much like the previous branch, this often involves leveraging the inherent processing capabilities of neuron-like units to perform some computational task which may itself be totally unrelated to biology. 07:31.820 --> 07:41.540 However, unlike neuromorphic computing, where we're using an engineered electronic device as our building block, under the in vitro approach we're utilizing actual neurons and interfacing these to external inputs and outputs. 07:41.860 --> 07:47.040 So rather than trying to recreate the behavior of a natural system, we're cutting out the middleman and harnessing the real thing. 07:47.320 --> 07:55.800 While plenty of research has been done on neuronal behavior in vitro over the past century, using neurons ex vivo as part of a larger non-biological system is a much newer idea. 07:56.240 --> 08:05.900 One of the challenges inherent in this approach is that unlike most circuits, neurons are living biological entities with constant metabolic requirements and a propensity to change in real time, sometimes in undesired ways. 08:05.920 --> 08:11.080 Short of freezing in sub-zero temperatures, there's no way to turn off a cell culture the way you can turn off an electronic device. 08:11.180 --> 08:14.380 That said, the biological nature of these systems can also be beneficial. 08:14.380 --> 08:22.400 These same inherent behaviors can offer the ability to self-organize, change in response to inputs, and theoretically heal to an extent impossible in electronic circuits. 08:22.700 --> 08:32.740 Finally, and of particular interest to the biohacker-turned-neuroengineer, cells can, with care, be grown and manipulated in a variety of settings, including those available to the enterprising biohacker. 08:32.800 --> 08:37.220 Certainly less stringent environments than the type of clean room required for microelectronic fabrication. 08:38.320 --> 08:42.920 So hopefully at this point, my method of delineating these different branches of research makes more sense. 08:42.960 --> 08:51.440 Between any two branches, there are a number of parallels that can be drawn, with variation only in either the end goal of the system or the exact strategy and methods used to accomplish that goal. 08:51.660 --> 08:58.460 It's also worth noting that, as defined, projects that fall into one of these branches have tended to remain within only one branch of neuroengineering. 08:58.520 --> 09:10.520 However, I would argue that as these technologies mature in these use cases to which they're applied grow in number, we'll see more efforts to join these approaches and leverage the unique strengths of each to allow us to do things we can only dream of at this point. 09:10.640 --> 09:15.460 This kind of synergistic approach is something that I'm personally interested in exploring in my own future research. 09:15.780 --> 09:20.020 So speaking of doing things, what are some actual examples of each of these types of neuroengineering? 09:20.020 --> 09:24.120 And as a reminder, this is intended to be a highlight reel and not a comprehensive overview. 09:24.300 --> 09:32.760 My goal is just to point out some specific efforts I think are of particular importance or interest, and I encourage you to use them as a starting point if you're interested in learning more about this kind of thing. 09:34.000 --> 09:41.320 So here's an example of a state-of-the-art neural interface array, which is the culmination of several years of development work by multiple collaborating labs. 09:41.640 --> 09:48.560 The Shepard and Viventi groups are both big players in the world of electrode arrays, and John Rogers is practically the founder of the field of soft electronics. 09:48.780 --> 09:52.720 I really encourage you to look in the work done by these labs if that kind of thing interests you. 09:52.960 --> 10:01.160 This device is an example of an electrode array that's similar in operating principle to an EEG array, but with some key differences that dramatically improve sensitivity and resolution. 10:01.380 --> 10:07.540 So this type of device is called an electrocorticogram, or ECoG array, and it's designed to be placed directly on the surface of the brain. 10:07.700 --> 10:16.160 Since neural activity recording resolution depends on both surface coverage and distance from the neurons being monitored, an array like this offers tremendous advantages. 10:16.260 --> 10:20.280 So first, the huge number of recording sites allow for high inherent spatial precision. 10:20.500 --> 10:27.360 Second, the thinness and flexibility of this array allows it to be molded to the curvature of the brain, improving coverage and minimizing separation distance. 10:27.360 --> 10:32.860 This is particularly important in the case of primates, and especially humans, where there's a large degree of surface contouring. 10:32.860 --> 10:38.880 So this design allows the array to be placed around the gyri and into the cell side, which would be impossible with a thicker array. 10:39.040 --> 10:51.780 Further, this array incorporates active electronics into the electrodes themselves, allowing for signals to be amplified near their origin and allowing multiple electrodes to be multiplexed between a smaller number of conductors running between the skull and the outside instrumentation. 10:51.920 --> 10:57.060 Without this type of signal management, high density arrays quickly become impossible to manage practically. 10:57.920 --> 11:01.740 And because it's sort of difficult to appreciate that scale, here's that array in higher resolution. 11:01.740 --> 11:07.840 So this is 1,008 electrodes in less than one square centimeter, which is pretty darn impressive. 11:09.060 --> 11:16.820 As mentioned previously, neuromorphic engineering applies biologically inspired processing to other computational problems, which might not have anything to do with biology at all. 11:16.820 --> 11:22.520 However, in this particular case, artificial neurons were used to model the activity of simulated portions of the brain. 11:22.800 --> 11:38.140 So due to the immense scale of connectivity in the brain, which is made up of about 86 billion neurons and vastly more synaptic connections, and the difficulty of accessing the living brain using instrumentation, simulation efforts like the Human Brain Project have played a vital role in the study of cognitive processing. 11:38.140 --> 11:43.300 That said, even these massive joint efforts are limited due to the sheer computational resources required. 11:43.320 --> 11:56.620 To help ameliorate this problem, researchers at the Human Brain Project leveraged the fact that artificial neurons fabricated from metallic and semiconducting materials have a huge intrinsic speed advantage over biological neurons due to differences in conduction mechanisms. 11:56.620 --> 12:06.560 So as a result, by simulating real neurons using engineered ones in silicon, limited neuronal processes were able to be simulated at a much faster rate than what's observed in real physiology. 12:06.560 --> 12:16.140 The brain scale system shown here utilizes a large number of artificial neurons fabricated on silicon wafers using conventional integrated circuit manufacturing techniques. 12:16.140 --> 12:25.360 These are created in bulk across the wafer surface, and unlike most integrated circuits, these are actually directly accessed from the intact wafer, so there's no dicing. 12:26.080 --> 12:28.940 So here's a closer look at the individual neuron circuits. 12:29.060 --> 12:35.540 At the bottom, that's a layout of two of the neural equivalent circuits that are fabricated in mass on that chip. 12:35.960 --> 12:47.540 They're arranged into one of 384 clusters of 512 neurons each onto a single wafer, and the assembly into which that wafer is loaded is shown on the right with supporting and interface electronics. 12:47.540 --> 13:09.660 So 20 wafer modules consisting of 384 clusters of 114,688 programmable synapses and 512 neurons each give a total of 880 million synapses and 3.9 million neurons for the entire system, which is further capable of running 100,000 times faster than the same set of neurons would in vivo. 13:09.660 --> 13:19.100 So these advantages and the ability to continuously monitor each simulated neuron in real time have enabled considerable advances in the field of computational neuroscience. 13:19.800 --> 13:27.320 So as I mentioned earlier, artificial hardware or software-based neurons aren't the only tools that neuroengineers use to solve processing tasks. 13:27.320 --> 13:30.700 So here's an example of real neurons being used to do the same thing. 13:30.700 --> 13:50.000 In 2005, researchers at the University of Florida were able to extract about 25,000 neurons from rat cortical tissue, seed them onto a 60-electrode microarray treated to function as a culture disk, and then observe the spontaneous formation of interneuronal connections and eventually synchronized firing activity around 10 days after initial seeding. 13:50.760 --> 14:00.800 Two electrodes were used for both stimulation and recording, and these outputs were fed to a flight simulator program where neuronal activity of these electrodes was used to establish pitch and yaw control. 14:00.800 --> 14:09.080 So by feeding back a stimulatory error signal based on telemetry output by the flight simulator, firing activity and neurons near the electrode were modified. 14:09.560 --> 14:20.560 And over the course of repeating the simulation multiple times, neuronal activity was found to change in self-weight in order to produce stable flight in response to changes in telemetry data over the course of about 15 minutes. 14:20.600 --> 14:33.340 So this feedback adjust and test mechanism is by itself very similar to techniques used to train software neural networks, which isn't surprising given the fact that these were in turn inspired by learning mechanisms seen in the nervous system. 14:33.340 --> 14:35.340 So it's come full circle here. 14:36.040 --> 14:45.200 So now that we've seen some examples of different types of neuroengineering research, I want to briefly discuss how biohackers might go about getting involved in this type of work on their own. 14:45.440 --> 14:48.740 So growing cells in general is not a difficult process. 14:48.740 --> 14:51.440 To quote Jeff Goldblum, life finds a way. 14:51.840 --> 14:58.360 The real difficulty in this process arises from the same principle, life finds a way, but it's not always the way that you want it to. 14:58.660 --> 15:09.080 So in order to successfully culture a specific cell type, you have to take careful consideration of the cell source, the conditions in which the cells are kept, and the intended final use of those cells. 15:09.200 --> 15:14.160 So regardless of type, though, a key principle of cell culture is aseptic technique. 15:14.320 --> 15:23.180 So care must be taken at all times in the process to avoid introducing other microorganisms into the culture environment, as these might outcompete or otherwise harm the cells of interest. 15:24.580 --> 15:29.040 So prior to culture, the cells must first be acquired from some suitable source. 15:29.040 --> 15:40.320 The simplest mechanism for obtaining neural cells is the harvest of primary cells from tissue samples, either acquired directly from the donor animal or ordered through a supplier such as BrainBits or some other vendor. 15:40.540 --> 15:44.640 The most common sources for these cells are rats, mice, and embryonic chicks. 15:45.280 --> 15:55.980 After harvest, the cells of interest have to be purified from this sample, which will also likely contain other cell types as well as other biomolecules and extracellular matrix proteins and connected tissue. 15:56.060 --> 16:05.540 In order to enrich the primary culture, these extra components have to be removed, typically with enzymes known as proteases, which break down matrix proteins, allowing the cells to be dissociated. 16:05.860 --> 16:11.900 Additional sorting based on cell type may also be needed to remove glial cells and other neuron phenotypes if necessary. 16:12.060 --> 16:19.020 So since mature neurons don't normally undergo division, primary cultures need to be reestablished periodically as more cells are required. 16:19.020 --> 16:30.860 To avoid this problem, another option is using immortalized secondary cultures, and these are derivatives of primary cells which have been modified to exhibit tumor-like behavior, including indefinite division. 16:31.080 --> 16:35.200 So these cells tend to take on very homogeneous characteristics, which can be beneficial. 16:35.280 --> 16:45.700 However, this modification is the possible downside of modifying the cellular characteristics in a way that adversely impacts the intended application, and it definitely can differ from the normal phenotype of the cell. 16:45.920 --> 16:48.560 So whether or not that's a deal breaker depends on the application. 16:49.080 --> 17:05.340 So a more complex third option is the use of progenitor cells, which are cells that are stem-like in that they're capable of undergoing both division and differentiation, which allows for a baseline stock of cells to be maintained while simultaneously producing an essentially indefinite supply of the final differentiated cells needed. 17:05.340 --> 17:17.820 So this route's tricky because the cells rely on specific and sometimes subtle external cues to guide their ultimate differentiation, which must be carefully monitored during culturing because it can change pretty rapidly. 17:18.080 --> 17:25.260 So one major advantage of this technique is that this represents one of the only ways of acquiring human neural cells. 17:25.260 --> 17:33.660 So mature cells can undergo induced dedifferentiation, which reverts them back into stem-like cells when appropriate genetic modifications are made. 17:33.700 --> 17:42.440 This procedure is moderately difficult, but it's fairly well established at this point and remains one of the only ways that you can get human neural cells as a culture, which is pretty cool. 17:43.120 --> 17:50.860 So in order to grow, cells have to be stormed in a warm, humid environment, and the supply of gas must be suitable for the type of cell being cultured. 17:50.860 --> 18:00.980 It's important for the environment to remain aseptic again, as it's essential for preventing other opportunistic organisms such as yeasts, bacteria, and viruses from colonizing the culture. 18:01.300 --> 18:07.380 Most abiding cell types will grow to the point where their density in the culture begins to negatively impact cell survival. 18:07.480 --> 18:17.180 When the cells reach a certain coverage of the culture chamber surface, termed confluence, some of the cells must be removed to prevent adverse effects from damaging the remaining cells. 18:17.420 --> 18:23.680 So this removal of cells can also be used to expand the culture back into additional flasks to build up stock if desired. 18:23.820 --> 18:25.760 This is a technique called passaging. 18:26.620 --> 18:33.560 And in addition to the different temperatures and atmospheric requirements , cells tend to differ even more with respect to their metabolic requirements. 18:33.560 --> 18:41.360 So fresh media has to be changed regularly and supplied appropriately, and it has to contain the required nutrients specific to that cell type. 18:41.700 --> 18:47.640 Further, neural cells are particularly sensitive to the choice of substrate or the material on which the cells directly rest. 18:47.820 --> 18:55.840 So frequently the culture chamber itself will be coated with an intermediate substance, often a polymer or substances normally found in the extracellular matrix like collagen. 18:56.080 --> 19:03.740 Different substrates can actually induce or prohibit different cell differentiation routes in some cases, or it may even cause de-differentiation. 19:03.740 --> 19:06.220 It entirely depends on the cell specific type. 19:06.380 --> 19:14.680 So certain substrates are able to quickly polymerize into 3D structures suitable for seeding cells into to form volumetric cultures. 19:14.720 --> 19:20.760 So this is sometimes used with neural progenitor cells to form 3D spatially differentiated structures called organoids. 19:22.120 --> 19:28.300 And then finally, you need to be able to interact with your cultured cells in a way that's appropriate for your specific application. 19:28.320 --> 19:36.380 So typically this is going to involve electrical or chemical detection or stimulation, usually referred to as electrophysiology and neurochemistry respectively. 19:36.400 --> 19:40.280 In some applications, both methodologies might be used at the same time. 19:40.280 --> 19:46.400 So groups of cells can be stimulated and local potentials can be measured using electrode arrays, which is those discussed earlier. 19:46.620 --> 19:52.580 If a neuron specific electrical activity needs to be measured, patch clamps and other similar devices might be used. 19:52.580 --> 19:57.880 In this technique, you take a glass capillary that's formed into a very small point that's filled with electrolyte. 19:57.880 --> 20:00.880 You then aspirate that against the membrane of a target cell. 20:00.920 --> 20:10.600 And what that does is it creates an electrical connection with the interior of the cell, allowing for the cell membrane potential to be measured or even driven with respect to the extracellular space. 20:11.000 --> 20:22.040 So neurochemical measurements, which are a little trickier, can be made using electrochemical techniques where specific analytes are oxidized or reduced against the surface of an electrode in response to a varying electrode potential. 20:22.340 --> 20:33.560 This works with only a subset of analytes, however, but there are additional tricks that can be used to increase selectivity or expand applicability to other molecules through the use of enzymes or other functional coatings. 20:34.640 --> 20:43.760 And then when the cells need to be arranged in specific ways, microfluidic devices can be created using soft lithography and seeded with cells, sort of like this one here. 20:44.060 --> 20:50.380 It's also entirely possible to create larger millifluidic devices that can be fabricated using fairly basic tools. 20:50.420 --> 20:52.440 I've seen 3D printed ones, for example. 20:52.840 --> 20:59.920 Generally, a positive pattern is made and an elastomer like PDMS or polydimethylsiloxane is poured onto the mold and then removed. 20:59.920 --> 21:05.500 And then that resulting cast is then bonded to a microscope slide using ozone to create a really tight seal. 21:05.920 --> 21:13.340 And then this can have fluid basically flowed through it with a watertight joint between the cast and the glass. 21:13.340 --> 21:19.560 And you can put cells in there and basically they're allowed to grow freely and then interact based on the geometry that's specified by the design. 21:20.720 --> 21:24.160 So to finish up, I have some useful links here. 21:24.160 --> 21:27.860 One of the benefits of doing this virtually is you all get access to the stuff after the fact. 21:27.860 --> 21:33.920 So if you're interested in suppliers of neural culture reagents and whatnot, BrainBits is a great resource for that. 21:34.040 --> 21:43.220 A bunch of the other vendors like Thermo Fisher have reagent and media guides and also tend to publish protocols for de-differentiation and differentiation of neural cells. 21:43.560 --> 21:47.760 So if you want to try getting started on any of this stuff at home, that would be a resource I would recommend. 21:47.960 --> 22:01.440 If you're interested in the Human Brain Project, which has really taken a lot of these methodologies and some of the systems that have been developed and applied them to problems of neuroscience, as I mentioned earlier, it sort of extends beyond neuroengineering, 22:01.440 --> 22:12.080 but I encourage you to check out the Human Brain Project in that case, and specifically their neuromorphic computing research platform, which I discussed earlier, of which the BrainScale system is just one facet. 22:12.420 --> 22:20.800 There's a bunch of other systems like that and other examples like IBM is coming out with some models of that on dedicated ICs currently. 22:21.140 --> 22:25.180 So there's a huge variety of different projects in all of these areas of neuroengineering. 22:25.260 --> 22:32.700 Like I said, this is just a very brief skim of some of the projects that I think are interesting and representative of each area. 22:33.200 --> 22:35.840 And so I definitely encourage you to check it out more if you're interested. 22:36.380 --> 22:39.620 With that, here's some other references as well. 22:39.900 --> 22:41.560 And thank you very much for the talk. 22:41.560 --> 22:44.020 I'd be happy to take any questions during the Q&A.